|
By Pam Fulmer
The call usually starts the same way. A general counsel describes a software implementation that consumed two years and eight figures and never worked. Then, before we get to the facts, she says the thing that almost stopped her from calling at all: “But I’ve read our agreement. Damages are capped at twelve months of fees. Consequential damages are waived. There’s an integration clause. We’re stuck.” She has read the contract correctly. She has drawn the wrong conclusion from it. A liability cap allocates the risk of a deal that goes badly. It is not a license to lie about what you are selling. Under California law, the vendor’s own paper is frequently the weakest part of its defense once the sales cycle is examined — and the clauses that look like a wall are, in a fraud case, considerably closer to a fence. Under California law, a limitation-of-liability provision remains enforceable for an ordinary contract claim in many commercial settings. But it cannot be used to limit liability for an adequately pleaded, independently tortious fraud or other willful injury. The distinction is important—and fact intensive. Civil Code section 1668 reaches damage limitations for willful torts California Civil Code section 1668 makes contracts that directly or indirectly exempt a party from responsibility for its own fraud, willful injury to another’s person or property, or violation of law contrary to public policy. In New England Country Foods, LLC v. VanLaw Food Products, Inc., 17 Cal.5th 703 (2025), the California Supreme Court held that section 1668 invalidates contractual limitations on damages for willful injury; the statute is not confined to provisions that eliminate every possible remedy. The Court also rejected a rule turning enforceability on the commercial sophistication of the parties or their bargaining power. That holding has an essential limit. Section 1668 does not invalidate an agreed limitation for a pure breach of contract merely because the breach was deliberate, opportunistic, or economically harmful. The claimant must establish a tortious wrong that is independent of the contractual nonperformance and falls within the statute’s scope. VanLaw did not decide whether the pleaded tort claims in that case actually met that standard. Fraud must be more than a disappointed implementation The economic-loss rule and the independent-duty doctrine remain central. A customer cannot obtain uncapped tort remedies simply by relabeling a failure to perform as “fraud.” In Robinson Helicopter Co., Inc. v. Dana Corp., 34 Cal.4th 979 (2004), the Court allowed a fraud claim based on affirmative, knowingly false certificates of conformance that were separate from the defendant’s delivery of nonconforming goods. The decision is deliberately narrow; it does not make every false assurance of contractual performance actionable in tort. More recently, Rattagan v. Uber Technologies, Inc., 17 Cal.5th 1 (2024), held that fraudulent concealment during a contractual relationship may support a tort claim only when the claim’s elements can be established independently of the parties’ contractual rights and obligations and the conduct exposed the claimant to a risk of harm beyond the parties’ reasonable contemplation when they contracted. For a software dispute, the strongest fraud facts usually arise from the sales cycle: a statement of present capability, a concealed material limitation, manipulated proof-of-concept results, or a promise made without a present intent to perform. The facts must show not merely that the product failed, but that the vendor deceived the customer to obtain the deal or inflicted an independent tort in performing it. An integration clause does not bar proof of actual fraud—but reliance still matters An integration clause does not categorically exclude evidence offered to prove that an agreement was procured by fraud. In Riverisland Cold Storage, Inc. v. Fresno–Madera Production Credit Ass’n, 55 Cal.4th 1169 (2013), the California Supreme Court overruled the restrictive Pendergrass rule and reaffirmed that the fraud exception to the parol-evidence rule may permit evidence of fraudulent promises even when they conflict with the written agreement. That does not make every precontract statement actionable. The claimant must still plead and prove a material misrepresentation or concealment, scienter, intent to induce reliance, justifiable reliance, and resulting damage. A sophisticated buyer’s diligence, the contract’s express warnings or disclosures, and any specific non-reliance provision can materially affect the reliance analysis. Remedies require a claim-by-claim analysis A viable, independent fraud claim can materially alter the remedies analysis, but the available remedy depends on the transaction and the facts. • Compensatory damages. For fraud in the purchase, sale, or exchange of property, Civil Code section 3343 generally provides an out-of-pocket measure and specified additional damages. It is not the universal measure for every software-services fraud claim. Other fraud claims may invoke the general tort measure, and contract and tort damages cannot duplicate the same loss. • Punitive damages.Civil Code section 3294 permits exemplary damages for a noncontractual obligation upon clear and convincing proof of oppression, fraud, or malice. For a corporate defendant, the statute requires the requisite conduct, authorization, or ratification by an officer, director, or managing agent. • Rescission and restitution. A party whose consent was obtained through fraud may have a right to rescind under Civil Code section 1689(b)(1). Rescission is an election of remedy with procedural and restoration obligations; it is not automatically available simply because an implementation was unsuccessful. • Limitations. A California fraud claim is generally subject to the three-year period in Code of Civil Procedure section 338(d), which accrues on discovery of the facts constituting the fraud. Discovery is not an indefinite extension; a claimant’s actual or inquiry notice may be contested. What to do before signing the next document Preserve the sales cycle. Collect decks, RFP responses, proof-of-concept scripts and outputs, product demonstrations, emails, chat messages, recordings, calendar invitations, and relevant internal notes. Preserve the sources and metadata, not just selected excerpts. Assess new paper carefully. A renewal, change order, remediation plan, release, or settlement may affect claims and remedies. Review it before signing; do not assume it merely addresses operational issues. Build the fraud record promptly. Document who made each representation, their role and authority, the recipient, date, medium, precise substance, contemporaneous evidence of falsity, reliance, and resulting harm. California fraud must be pleaded with specificity. Lazar v. Superior Court, 12 Cal.4th 631 (1996), explains that a corporate fraud allegation ordinarily must identify the speakers, their authority, what they said or wrote, to whom, and when. The practical point A liability cap may determine the value of an ordinary contract dispute. It does not resolve whether the vendor made an independently actionable misrepresentation, concealed material facts in circumstances giving rise to a duty, or engaged in another willful tort. Those questions require a careful examination of the sales process, the contract language, the nature of the alleged wrong, and the damages actually caused.
0 Comments
A Tactical Take on the gap between what’s promised and what’s signed
If your company is evaluating a new ERP system, the sales process can feel like the most attentive, professional courtship you’ve ever experienced. That’s by design. Understanding how the cycle actually works — and where the traps are — can save you from signing a contract that bears little resemblance to what you were sold. The Charm Offensive It usually starts with an impressive show of force. The vendor trots out a small army from their side to “learn your business.” There are discovery calls, scoping sessions, and workshops — a steady rhythm of Zoom, Teams, and in-person meetings, each one devoted to understanding how your operation runs. It looks like a serious commitment of resources, and it feels flattering. Woven through all of it is the language of collaboration. The sales team talks about wanting to partner with you, to be a long-term ally rather than just a vendor. It’s warm, it’s reassuring, and it’s meant to lower your guard. Here’s what to keep in mind: the people in those rooms are often genuinely eager but not always genuinely informed. An enthusiastic salesperson may know very little about the technical realities of the product, yet will confidently represent that the software can do everything your company needs. Enthusiasm is not the same as a commitment, and a friendly assurance is not the same as a specification. Everything is Oral — and That’s the Point Notice how much of the scoping conversation happens out loud and how little of it lands in writing. Promises about capabilities, custom workflows, integrations, and outcomes tend to be made verbally, across dozens of meetings, with very little committed to paper. That is not an accident. The fewer written representations there are, the smaller the paper trail — and the less the vendor can be held to later. The Pricing “Test” and the Expiring Discount Once you’re hooked, the pricing arrives. Treat the first number as what it is: a test of how much they can get. When you hesitate, the discounts appear — often steep ones — accompanied by urgency. You’ll be told the discount is expiring, that you need to sign now, and maybe that this is the best pricing the rep has ever seen. It usually isn’t. Manufactured urgency and “best I’ve ever seen” framing are standard closing tools, not reflections of a genuinely rare opportunity. A real deal is still a real deal next week. The Paperwork Switcheroo Once you agree to move forward, the paperwork comes over — frequently through DocuSign, ready for a quick signature. This is the moment to slow down, because this is where the gap opens up. The specifics you spent weeks scoping are often nowhere in the agreement. Instead of the tailored solution you discussed, you get a very general scope of work that promises little more than the vendor’s generic, off-the-shelf product. That’s because once you are in implementation and you ask why the software does not have the required functionality you discussed, the ERP vendor pulls out the contract and says it was never included. Then there’s the order form — often a mystery. It lists multiple modules that, the publisher assures you, will combine to deliver your solution. What’s frequently missing are the key pieces that actually make it work: connectors, integrations, and other essential components. They’re sometimes left out precisely because the sales team fears that including them — and their cost — will tank the deal. After signing, these reappear as change orders or as required add-on items, at an additional expense. The clause that erases every promise Buried in the contract is an integration clause (also called a “merger” or “entire agreement” clause). In plain terms, it states that the signed written contract is the complete and final agreement between the parties, and that it supersedes every prior discussion, representation, and promise made during the sales process. Read that again in light of everything above. All those oral assurances about what the software could do? Legally, they’re gone. If it isn’t in the written contract, you generally can’t rely on it. The integration clause is the mechanism that turns weeks of verbal promises into thin air the moment you sign. The Disappearing Act Then the deal closes, and the dynamic changes overnight. The attentive sales team is gone with the wind. In their place arrives the implementation team — and this is often the moment the customer starts learning that what was promised during the sales cycle cannot actually be done, at least not without more time, more modules and add-ons, and more money. What to Watch For
The sales cycle is engineered to feel like a partnership. The contract is what actually governs the relationship. Make sure the two agree before you sign. By Pam Fulmer
The email arrives on a Tuesday afternoon. It is polite. It is addressed to someone in procurement or IT who has never seen one before. It references a section number in an agreement signed years ago by people who no longer work at the company. It asks for a call to “discuss your licensing position” and requests that you run an attached script and return the output within thirty days. There is no dollar figure in it. That comes later — and by then, the number will have been built almost entirely out of information your own team handed over in the first week. This is the central and counterintuitive fact about software license audits: the most consequential decisions are made before anyone knows what is at stake. By the time a seven- or eight-figure demand arrives, the evidentiary record supporting it is already closed, and your company built most of it. The audit demand is not calculated from what you deployed. It is calculated from what you reported, measured the way the publisher's tools measure it. You are not being singled out Clients often assume an audit letter means the publisher suspects something. Usually it does not. Audits are a revenue channel, run on a schedule, and the volume is substantial. Flexera's 2026 State of ITAM Report, drawn from a survey of ITAM professionals conducted in early 2026, found that nearly half of organizations — 48 percent — received a software audit in the prior twelve months. Microsoft was the most frequently reported auditor at 64 percent of those audited, with Oracle close behind at 50 percent. Oracle's reported audit activity jumped roughly fourteen percentage points year over year. Adobe and VMware activity also rose. The cost side is where it stops being a compliance annoyance and becomes a balance-sheet event. Forty-four percent of organizations reported spending more than $1 million on software audits over a three-year period. In earlier Flexera data, the share of companies reporting audit costs above $10 million nearly doubled year over year, from 7 percent to 12 percent. Two conclusions follow. First, receiving a letter is not evidence that you did anything wrong. Second, the process that follows is expensive enough that it deserves the same seriousness as any other bet-the-budget dispute — from hour one. Hours 0 to 4: contain the response The single most damaging thing that happens in a software audit usually happens in the first afternoon, and it is almost always well-intentioned. A systems administrator receives the letter, wants to be helpful and cooperative, and replies directly to the auditor. In doing so, he or she confirms deployment figures from memory, describes the virtualization architecture, explains that a subsidiary has been using the software too, or apologizes for something. None of it is verified. All of it is admissible. All of it is now the publisher's baseline. What to do instead
Hours 4 to 24: preserve everything, change nothing Two instincts surface once the seriousness registers, and both are dangerous. The first is to clean up — uninstall software that may be over-deployed, decommission instances, tidy the environment before anyone looks. Do not. Depending on the contract and the circumstances, this can convert a commercial dispute into an allegation of spoliation or bad faith, and publishers have begun litigating precisely this point. In the Broadcom litigation now pending in the Northern District of California, the publisher's position is essentially that a customer cannot extinguish audit obligations by removing the software mid-review. Whatever the outcome, you do not want to be the test case. The second is to delete — to purge old email threads about license planning or the failed true-up conversation from two years ago. That is worse. What to do instead
Hours 24 to 48: read the contract, not the letter The audit letter describes what the publisher would like to happen. The contract describes what the publisher is entitled to make happen. In our experience these two documents diverge more often than they align, and the gap is where the leverage is. Working through the actual audit clause, ask:
Hours 48 to 72: build your own number first An audit is a valuation dispute conducted under the appearance of a factual inquiry. Whoever establishes the measurement framework first usually wins the argument that follows. Before responding substantively, and under privilege, develop an independent internal picture: what is actually deployed, what entitlements actually exist, and what the defensible license position looks like under a reasonable reading of the contract. This accomplishes three things. It tells your leadership the realistic range of exposure rather than the publisher's opening number. It surfaces the contractual ambiguities worth fighting over. And it means that when the publisher's figure arrives, you are comparing two analyses rather than reacting to one. Where the technical measurement is genuinely complex — virtualization, containerization, indirect or digital access, disaster-recovery environments, development and test instances — a licensing specialist working at counsel's direction is usually worth the cost several times over. The distinction matters: retained through counsel, that work carries a privilege argument. Retained directly, it generally does not. The five mistakes that cost the most money 1. Treating it as an IT problemAudits are contract disputes that happen to involve technology. Routing one to the infrastructure team because it mentions servers is how companies end up bound by admissions no lawyer ever reviewed. 2. Volunteering scopeAnswering questions that were not asked, disclosing affiliates the publisher had not identified, and describing environments outside the audited product family all expand the audit at your own expense. 3. Accepting the publisher's mathDeployment counts under contested interpretations — core-counting in virtualized environments, user definitions that sweep in service accounts, indirect access theories — are legal conclusions dressed as measurements. They can be contested. They frequently should be. 4. Negotiating on the publisher's clockAudit findings tend to arrive near a quarter or fiscal year end, accompanied by a deadline and a discount that expires. The urgency is manufactured, and the resolution offered is almost always a purchase rather than a settlement — the compliance gap is forgiven in exchange for a subscription commitment considerably larger than the gap. 5. Waiting to involve counsel until there is a demandBy then the record is fixed. Counsel engaged at the demand stage is negotiating over evidence the company created without advice. Counsel engaged at the letter stage is shaping what that evidence is. What resolution actually looks like Most audits do not end in litigation. They end in a negotiated commercial resolution, and the quality of that resolution depends almost entirely on the strength of the customer's independent position and its credible willingness to litigate if pushed. A well-defended audit typically produces some combination of a substantially reduced compliance figure, a release covering the audited period, an agreed and documented licensing interpretation going forward, amended audit provisions with tighter scope and notice requirements, and — where the publisher's conduct warrants it — a negotiated standstill on further audits for a defined period. What it does not produce is a rushed subscription commitment signed three days before the publisher's fiscal year closes. When you need litigation counsel and not just a licensing consultant Licensing consultants are valuable, and we work with them regularly. But certain features of an audit signal that the matter has already left the commercial track:
If an audit letter has arrived Our practice is devoted to software licensing and technology disputes — audit defense, over-deployment and indirect access claims, failed implementations, and litigation against major publishers. If your company has received an audit notice, or expects one, the most valuable conversation is the one that happens before the first substantive response goes out. This article is provided for general informational purposes and does not constitute legal advice or create an attorney-client relationship. By Pam Fulmer
Our earlier piece made two points that sit in tension. First, when a software vendor’s exclusive “repair or replace” remedy collapses, Article 2 of the Uniform Commercial Code may give a buyer a statutory path around contractual remedy limitations. See Cal. Com. Code § 2719. Second—and inconveniently—that doctrine applies only if Article 2 governs the transaction. A modern cloud ERP delivered as software-as-a-service may be characterized as a service rather than a sale of goods. If a court holds that the enterprise system was a service, the UCC’s cleanest tool may disappear with it. So, the natural question is: if the UCC is off the table, is the customer defenseless against a vendor’s one-sided agreement? No. California common law and the Civil Code supply a different set of tools. A 2025 California Supreme Court decision, New Eng. Country Foods v. Vanlaw Food Prod., Inc., 567 P.3d 63, 331 Cal. Rptr. 3d 890 (Cal. 2025), has made one of them--Civil Code Section 1668—considerably sharper. First, an Honest Concession There is no freestanding California common-law doctrine called “failure of essential purpose.” That phrase belongs to UCC Section 2-719 and its California counterpart, Commercial Code Section 2719. A customer cannot simply invoke it in a services case and expect the same statutory framework to apply. What California does have is a cluster of common-law and statutory doctrines that can accomplish some of the same practical work: freeing a customer from contractual limitations when the vendor’s conduct, or the collapse of the bargain, justifies relief outside ordinary capped contract damages. The key is to understand which tool gets you past the vendor’s damages cap. Ordinary breach doctrine may establish liability, but it often leaves a negotiated limitation-of-liability clause standing. Rescission, Civil Code section 1668, fraud, and unconscionability are the doctrines that attack the limitations themselves. Tool One: Rescission for Failure of Consideration The closest functional cousin to “you are not trapped by a remedy that failed” is rescission. A contract is extinguished by rescission. Cal. Civ. Code § 1688. A party may rescind where the consideration for its obligation fails, in whole or in part, through the fault of the other party, or where the consideration fails in a material respect before it is rendered. Cal. Civ. Code § 1689(b)(2), (b)(4). For a failed ERP, the theory is intuitive. The customer bargained for an operational, integrated system that would run core business functions. If the vendor never delivers a working system—go-live never occurs, the system is rolled back, or critical functions never work—the customer may argue that the consideration failed in a material respect. The strategic payoff can be significant. Rescission seeks to unwind the agreement rather than enforce it. If successful, the customer may obtain restitutionary relief and avoid contractual limitations that presuppose an enforceable contract. See Cal. Civ. Code § 1692. California courts recognize that rescission may support broader restorative relief designed to return the parties to their pre-contract positions. See Runyan v. Pacific Air Industries, Inc., 2 Cal. 3d 304 (1970). But the tradeoffs are real. Rescission requires prompt notice after discovering the facts entitling the party to rescind and generally requires restoration, or an offer to restore, what the rescinding party received. Cal. Civ. Code § 1691. It is also a different remedial posture from expectation damages: rescission is about unwinding the deal, not keeping the contract and recovering the full benefit of the bargain. For a customer that has sunk years and millions into a system it still wants fixed, that is a meaningful choice. Pleading rescission and damages in the alternative can preserve flexibility. Tool Two: Civil Code § 1668—and the 2025 VanLaw Decision The vendor’s most powerful weapons in a failed-implementation case are often its limitation-of-liability clause and consequential-damages waiver. California Civil Code Section 1668 is the most direct statutory answer to those provisions when the claim involves fraud, willful injury, or violation of law. Section 1668 provides that contracts having as their object, directly or indirectly, exemption from responsibility for one’s own fraud, willful injury to the person or property of another, or violation of law are against public policy. Cal. Civ. Code § 1668. Two California Supreme Court decisions are especially important. First, in City of Santa Barbara v. Superior Court, 41 Cal. 4th 747 (2007), the court held that an agreement purporting to release liability for future gross negligence is generally unenforceable as a matter of public policy. That rule does not depend on the transaction-by-transaction public-interest analysis used for ordinary negligence releases. Second, in New Eng. Country Foods v. Vanlaw Food Prod., Inc., 567 P.3d 63, 331 Cal. Rptr. 3d 890 (Cal. 2025), the California Supreme Court held that Section 1668 invalidates limitations on damages for willful injury to the person or property of another. The court rejected the argument that Section 1668 reaches only complete exculpation clauses and not damages limitations. It also rejected a case-by-case exception based on the parties’ sophistication or commercial bargaining context. VanLaw matters because many vendor agreements do not say, “we are not liable for intentional wrongdoing.” They instead cap damages, exclude lost profits, and bar consequential or punitive damages. VanLaw confirms that, at least for willful injury within section 1668, a clause that substantially limits damages can be invalid even if it does not eliminate every theoretical remedy. The implications for ERP litigation are substantial. If the customer can establish a qualifying independent tort—such as intentional misrepresentation, intentional interference, or other willful injury—or a qualifying violation of law, the vendor’s damages cap and consequential-damages waiver may be unenforceable as to that claim. That result is UCC-independent: it does not matter whether the ERP is classified as a good or a service. But the limitation is equally important. Section 1668 does not turn an ordinary breach of contract into an uncapped tort case. VanLaw expressly preserved the distinction between breach of contract and violation of an independent duty. The court stated that Section 1668 does not preclude parties from limiting liability for pure breaches of contract absent violation of an independent duty within the statute’s scope. Recent California authority also continues to police that boundary through the economic loss rule. See Rattagan v. Uber Technologies, Inc., 553 P.3d 1213, 324 Cal. Rptr. 3d 433 (Cal. 2024); Robinson Helicopter Co., Inc. v. Dana Corp., 34 Cal. 4th 979 (2004). The practical upshot is that the Section 1668 route demands real tort or statutory-duty facts. A disappointing system is not enough. A concrete pre-sale misrepresentation, concealment, willful injury, gross negligence, or qualifying statutory violation may be. Tool Three: Fraudulent Inducement Fraud is both a claim in its own right and a gateway to the tools above. ERP sales cycles are full of representations: capability demos, “it does that out of the box,” implementation timelines, integration promises, industry-fit assurances, and statements about the vendor’s experience with comparable deployments. Where those representations were false, material, and relied upon, fraudulent inducement can support rescission and can also bring section 1668 into play. The fraud claim must be pleaded and proved with discipline: the specific statement, the speaker, the timing, the falsity, the customer’s reliance, and resulting harm. California law supports the general principle that limitation-of-liability clauses do not protect fraud. In Food Safety Net Servs. v. ECO Safe Sys. USA, Inc., 209 Cal. App. 4th 1118 (2012), the court stated that limitation-of-liability clauses are ineffective as to fraud and misrepresentation under section 1668. But Food Safety is also a cautionary case: the fraud claim failed because the plaintiff did not establish tortious conduct independent of the contract, and the economic loss rule barred repackaging contract nonperformance as fraud. That is the lesson for ERP disputes. Fraud framed as generalized disappointment will not survive. Fraud tied to concrete, provable, extra-contractual or pre-contractual misrepresentations may unlock rescission, section 1668, and uncapped tort remedies. Tool Four: Unconscionability Under Civil Code § 1670.5 Unconscionability is not a UCC-only concept. California’s general unconscionability statute empowers courts to refuse to enforce an unconscionable contract or clause, or to limit its application to avoid an unconscionable result. Cal. Civ. Code § 1670.5. The familiar framework has both procedural and substantive components. Procedural unconscionability focuses on oppression or surprise, often from unequal bargaining power or non-negotiable form terms. Substantive unconscionability focuses on overly harsh or one-sided terms. A limitation clause that leaves the customer with no meaningful remedy for a total implementation failure may be a candidate for substantive unconscionability. The caveat is that unconscionability is an uphill fight between sophisticated commercial parties negotiating a major enterprise deal. California courts generally enforce limitation-of-liability clauses in ordinary commercial contracts unless a doctrine such as unconscionability, section 1668, or public policy applies. Food Safety illustrates that point: the court enforced a limitation clause against contract and ordinary-negligence theories where the plaintiff did not show unconscionability or a public-interest basis to invalidate it. Unconscionability is therefore best treated as a supporting theory, especially where the vendor’s terms were genuinely non-negotiable, the cap is illusory, or the remedy structure leaves the customer with no practical recourse for catastrophic failure. Tool Five: Material Breach—and Why It Is Not Enough on Its Own Common law has its own concept of a failure so serious that it excuses further performance: material breach. In a failed ERP case, the customer may argue that the vendor’s failure to deliver a functioning system was a material breach that discharged the customer’s remaining obligations and supported damages. But this doctrine does not, by itself, defeat a limitation-of-liability clause. Limitation clauses are drafted for breaches; that is their purpose. California courts generally enforce them in ordinary commercial settings unless a separate doctrine attacks the clause itself. That is why material breach supplies the liability theory, but rescission, section 1668, fraud, and unconscionability do the cap-defeating work. A customer should not assume that proving a catastrophic breach automatically means recovering uncapped consequential damages. The Balance Sheet Compared with the UCC path, the common-law playbook has disadvantages worth stating plainly. There are no UCC implied warranties if Article 2 does not apply. The customer must rely on express contractual promises, tort duties, statutory doctrines, and equitable remedies. Section 1668 requires more than a breach; it requires fraud, willful injury, violation of law, or another qualifying independent duty. The economic loss rule limits attempts to repackage disappointed contractual expectations as tort claims. Rescission requires prompt action and may force a strategic election. Unconscionability is difficult between sophisticated parties. But the advantages are real. These doctrines do not depend on proving that a cloud ERP is a “good.” Rescission can unwind the contract. Section 1668 can invalidate damages limitations for qualifying fraud, willful injury, or statutory-duty claims. And after VanLaw, a vendor cannot save a damages cap merely by arguing that the clause leaves some theoretical remedy in place. A Practical Playbook Plead in the Alternative and Preserve Rescission Early Preserve breach-of-contract damages, but plead rescission promptly where the facts support a material failure of consideration. Delay can waive rescission, so the remedy should not be an afterthought. Build the Fraud Record From Day One Collect demos, RFP responses, implementation promises, capability matrices, sales emails, and executive assurances. The question is not just whether the project failed; it is whether the vendor made a specific false representation or concealment that induced the deal. Frame Independent Torts Carefully To survive the economic loss rule, identify duties and conduct independent of the contract. Do not rely on artful relabeling of missed milestones or defective performance. Tie tort claims to misrepresentation, concealment, willful injury, or other conduct recognized as independent under California law. Use Section 1668 Where the Claim Qualifies After VanLaw, Section 1668 should be front and center when the customer has a genuine intentional tort or willful-injury theory. It should not be overused for ordinary breach. Its power comes from matching the doctrine to the facts. Negotiate Express Warranties on the Front End Because implied UCC warranties may not apply, buyers should negotiate express warranties and objective commitments: acceptance criteria, go-live conditions, data-migration accuracy, integration requirements, uptime, response times, project staffing, and remedies that are meaningful even if Article 2 never applies. Watch the Choice-of-Law Clause California’s Civil Code Section 1668 and VanLaw are favorable to customers with qualifying tort or statutory-duty claims. A vendor’s out-of-state governing-law clause may materially change the analysis. Bottom Line The UCC gives a wronged software buyer one clean, well-worn statutory doctrine. California common law and the Civil Code give the buyer a more complex toolkit: rescission, Section 1668, fraud, unconscionability, and material breach.. For SaaS deals that dominate enterprise buying today, that distinction is no longer academic. And after VanLaw, a vendor’s limitation clause is more vulnerable when the customer can plead and prove genuine intentional wrongdoing or another independent duty within section 1668. The customer whose implementation was sold on promises the vendor could not—or never intended to—keep may have a viable path outside the UCC. But the path must be built deliberately, with facts that attack the limitation clause itself rather than merely proving that the project failed. This article is provided for general informational purposes only and does not constitute legal advice or create an attorney-client relationship. The doctrines discussed are fact-intensive and continue to develop. The availability of rescission, the reach of Civil Code section 1668 after VanLaw, and the operation of the economic loss rule all turn on the specific contract and conduct at issue. Companies facing a cloud ERP dispute should consult qualified counsel. By Pam Fulmer
Is Your Cloud ERP Even Governed by the UCC? When an enterprise software implementation goes sideways, a customer’s lawyers often reach for the Uniform Commercial Code. Implied warranties of merchantability and fitness. Perfect tender. And the failed-remedy rule: when an exclusive or limited remedy fails of its essential purpose, the buyer may be able to reach the Code’s default remedies. See Cal. Com. Code § 2719; RRX Industries, Inc. v. Lab-Con, Inc., 772 F.2d 543 (9th Cir. 1985). There is a threshold problem that a surprising number of complaints skate past: those tools live in Article 2 of the UCC, and Article 2 governs transactions in goods. For delivered, licensed, on-premise software of the 1990s and 2000s, that was a fight worth having and often winnable. For the cloud ERP that dominates enterprise buying today—Oracle Fusion, Workday, SAP S/4HANA Cloud, NetSuite, Dynamics 365—the answer is far less certain and may well be no. A customer who builds its entire case on the UCC without first confirming that the UCC applies is building on sand. This piece explains why the goods-versus-services question has become central in modern ERP disputes, why software-as-a-service (SaaS) is hard to fit inside Article 2, and what sophisticated buyers should do about it. Why Classification Can Decide the Case The stakes of classification are not academic. If a transaction is governed by Article 2, the buyer may have access to the UCC’s buyer-protective architecture: implied warranties that attach unless effectively disclaimed, a demanding tender standard, and statutory remedies when an agreed remedy fails. If the transaction is governed instead by common law service-contract principles, the buyer may face a different regime: no implied warranty of merchantability, no Article 2 perfect-tender rule, and no statutory failure-of-essential-purpose provision. That does not mean common law leaves the buyer without remedies. It does mean the buyer’s path is different, and the vendor’s carefully drafted limitations may be harder to dislodge. The same failed implementation can therefore produce very different outcomes depending on a classification question the parties may never have consciously negotiated. That is why cloud vendors increasingly draft around the issue. Their forms often characterize the offering as “services,” confirm that the provider retains ownership of the software, and state that the customer receives only access rights—not delivery, title, or possession of a copy. Those labels are not automatically controlling, but they are important evidence of the transaction the parties made. The Framework: Goods, Sales, and Predominant Purpose Two threshold requirements must be satisfied before Article 2 governs, and cloud software strains both. First, the subject matter must be goods—defined in California as “all things ... which are movable at the time of identification to the contract for sale.” Cal. Com. Code § 2105. Second, there must be a sale, which the Code defines as the passing of title from seller to buyer for a price. Cal. Com. Code § 2106. A transaction that confers only a right to access or use software, without transferring title or a copy, sits uneasily inside that definition. When a contract blends goods and services—as nearly every enterprise software deal does—courts generally ask which aspect predominates. The classic formulation asks whether the contract’s predominant factor, thrust, or purpose is the rendition of services with goods incidentally involved, or a transaction of sale with labor incidentally involved. Bonebrake v. Cox, 499 F.2d 951, 960 (8th Cir. 1974). California courts apply a similar essence-of-the-transaction analysis. See Ventures v. SVC–W., L.P., 202 Cal. App. 4th 1483, 1502–03 (2012). The classification inquiry is typically contract-wide. If services predominate, common law—not Article 2—will generally govern the contract, including the software component. Why SaaS Is Different From the Software the Case Law Was Built On Much of the software-as-a-good precedent comes from a world of delivered products. Courts held that software could be treated as a good where something was delivered or transferred: a disk, a copy, a bundled hardware-and-software system, or a standardized software product. The leading example is Advent Systems Ltd. v. Unisys Corp., 925 F.2d 670 (3d Cir. 1991), which held that software was a “good” under the UCC where the contract’s main objective was the transfer of software and related products, with training and support treated as ancillary. The Ninth Circuit reached a similar result in RRX Industries, Inc. v. Lab-Con, Inc., 772 F.2d 543 (9th Cir. 1985), holding that the sales aspect of a software-system transaction predominated and that employee training, repair services, and upgrades were incidental to the sale of the software package. RRX is also important for remedies: applying California Commercial Code section 2719, the court affirmed consequential damages where the seller’s default was sufficiently total and fundamental that the limitation failed. But RRX should not be overread. It involved an installed software system in the 1980s, not a modern cloud-access subscription. SaaS changes the premise. In a true cloud ERP subscription, the customer usually receives no disk, no download, no executable, no object code, and no title to a copy. The customer receives a time-limited right to access software running on the vendor’s infrastructure, often priced by users, modules, or consumption. When the subscription ends, access ends. That structure creates problems on both Article 2 prongs. On the “goods” prong, there may be no movable thing delivered to the customer. On the “sale” prong, there may be no transfer of title. Some courts have applied Article 2 to software licenses, but the more cloud-like the transaction becomes, the harder it is to describe it as a sale of goods. Marquette University v. Kuali, Inc., 584 F. Supp. 3d 720 (E.D. Wis. 2022), is the modern SaaS decision buyers and vendors both should know. Applying Wisconsin law, the court accepted that the software underlying Kuali Research Cloud could be a good, and it accepted for purposes of analysis that the agreement included a sale. But it held that the contract was predominantly for services because Marquette paid for hosted access, maintenance, backups, updates, support, and the infrastructure needed to use the software—not primarily for a transferred software copy. Because services predominated, the UCC did not apply, and the contract’s remedy and damages limitations controlled. That reasoning maps naturally onto many cloud ERP subscriptions. The customer may care about the software functionality, but what it buys is often the vendor’s continuing operation of a hosted environment: access, uptime, maintenance, security, updates, integrations, and support. Those are service-like features, not the transfer of a movable chattel. The Law Is Unsettled—and the Facts Matter None of this means the UCC argument is dead. The authority is mixed, and the outcome depends heavily on the structure of the transaction, the governing law, and the factual record. A buyer has its best Article 2 argument when the deal involves standardized software, a meaningful delivered or downloadable component, or a contract whose economic center is the transfer of software functionality rather than ongoing vendor labor. Cases such as Advent and RRX remain useful for that proposition. The vendor has its best common-law argument when the contract is framed as a subscription service, the customer receives access rather than possession, the vendor retains title and operational control, and the customer pays materially for hosting, maintenance, configuration, support, and implementation. Kuali is the strongest modern SaaS example. California-specific authority also supports a careful, fact-based approach. In Tk Power, Inc. v. Textron, Inc., 433 F. Supp. 2d 1058 (N.D. Cal. 2006), the court applied common law rather than the UCC to a transaction involving prototype development because the essence of the agreement was development work—knowledge, skill, and ability—not the sale of finished goods. The court contrasted RRX, where the software sale predominated, with transactions centered on custom development. Other courts have drawn similar lines in technology disputes. In Conwell v. Gray Loon Outdoor Marketing, 906 N.E.2d 805 (Ind. 2009), the Indiana Supreme Court held that the UCC did not govern a website-design and hosting relationship because the arrangement involved custom design and ongoing hosting services, not a conventional transaction in tangible goods. The better formulation, then, is not that “software is a good” or “SaaS is a service.” The better formulation is: delivered, standardized software is often treated as a good; custom development, hosted access, and SaaS subscriptions often create stronger service-contract arguments. The Implementation Layer Makes the Goods Argument Harder Modern ERP deals rarely stop at a subscription. They often bundle implementation: configuration, data migration, integration, testing, change management, and training. These services may be performed by the vendor, a systems integrator, or both, and they may cost as much as—or more than—the subscription itself. Under the predominant-purpose test, that implementation layer matters. The more the buyer pays for the vendor’s labor, expertise, configuration, and project execution, the easier it is for the vendor to argue that services predominate. The irony is sharp: the more comprehensive and hands-on the vendor’s involvement—the very thing a customer wants when buying a mission-critical system—the harder it may become to characterize the transaction as the acquisition of a good. A deal in which the customer receives no delivered software copy, no title, and substantial implementation labor is difficult to shoehorn into Article 2. That does not make the UCC argument impossible. It does mean the argument must be pleaded and supported deliberately. What Sophisticated Buyers Should Do The practical response is not to abandon UCC theories. It is to stop depending on them and build a case that survives either classification. Plead in the Alternative Preserve the goods argument where the facts support it—especially for standardized suites, delivered components, downloadable modules, or arrangements where software functionality is the economic center of the deal. But assume the vendor will argue that the transaction is a service, and make sure the complaint survives if the court agrees. Build UCC-Independent Theories From Day One Fraud, negligent misrepresentation, and other pre-contractual representation theories do not depend on Article 2. ERP sales cycles often include capability representations, implementation assurances, timeline commitments, and integration promises that may support claims independent of the UCC. In California, Civil Code § 1668 provides that contracts that exempt a party from responsibility for its own fraud, willful injury, or violation of law are against public policy. That statute can be important when a vendor invokes limitation-of-liability language against fraud-based claims, regardless of whether the transaction is classified as goods or services. Negotiate Express Warranties Because Implied Ones May Never Attach If the transaction is classified as a service, implied UCC warranties may not apply. Buyers should negotiate explicit contractual warranties instead: implementation milestones, defined acceptance criteria, uptime commitments, integration obligations, performance standards, data-migration requirements, and meaningful remedies if the system does not work. Draft Remedies That Work Outside the UCC Do not rely solely on section 2719. A buyer should negotiate remedies that are enforceable as contract terms even if Article 2 never applies: refund rights, service credits that are not the exclusive remedy, termination rights, milestone holdbacks, fee clawbacks, audit rights, and carveouts from damages exclusions for mission-critical failures, data loss, security breaches, fraud, willful misconduct, and confidentiality breaches. Mind Choice-of-Law and Forum Clauses The goods-versus-services question can turn on governing law. Some jurisdictions are more receptive than others to treating software as a good; others focus more heavily on hosting, customization, and services. A vendor’s governing-law clause may therefore shape the classification fight before the dispute begins. Read the Vendor’s Characterization—and Rebut It Deliberately If the agreement says the offering is a “service,” states that the provider retains all rights in the software, disclaims delivery of any copy, and limits the customer to access rights, expect the vendor to rely on that language. If the buyer intends to invoke Article 2, it should develop a record showing why the transaction’s substance was the acquisition of software functionality rather than the purchase of labor or hosted operations. Bottom Line For a generation, “the software failed, so the UCC lets us past the vendor’s limitations” was a powerful opening move. In the cloud era, that move depends on a threshold question with an increasingly uncomfortable answer: a pure SaaS ERP subscription, especially one bundled with substantial vendor-led implementation, may fall outside Article 2 entirely. The sophisticated posture is not to assume the UCC applies, nor to concede that it does not. It is to recognize that the question is unsettled, litigate it deliberately where the facts support it, and—above all—build a case that does not collapse if the court decides the enterprise system was a service all along. This article is provided for general informational purposes only and does not constitute legal advice or create an attorney-client relationship. The classification of cloud and SaaS transactions under the UCC remains fact- and jurisdiction-specific. Companies facing a cloud software dispute should consult qualified counsel about their particular contract and circumstances. Breaking the Cap: Fraud in the Inducement in Failed Oracle Fusion and other ERP Implementations7/11/2026 By Pam Fulmer When a nine-figure ERP transformation collapses, the vendor’s limitation-of-liability clause is usually the whole ballgame. Here is how a well-pleaded fraud claim can get you past it — under California law, and in the two other jurisdictions where these disputes often land. Every large enterprise software dispute eventually comes down to the same question, and it is rarely the one the client expects. The client wants to talk about whether Oracle Fusion Cloud actually delivered the manufacturing or payroll functionality that was promised, or whether SAP’s S/4HANA rollout ever came close to going live. Those are the merits. But the number that decides whether the case is worth bringing — and whether it settles for real money — is buried in a boilerplate paragraph near the back of the Master Agreement: the limitation of liability. In a failed $30 million transformation, the cap is the ballgame. If the vendor’s limitation clause holds, the customer’s recovery is contractually squeezed down to the fees paid under the order form, with all consequential and “indirect” damages waived. The wasted internal labor, the second implementation, the lost efficiencies, the emergency consultants brought in to keep the lights on — none of it counts. A customer that spent $18 million and got nothing may find its recovery capped at a few million, minus what it still “owes.” The single most effective way past that cap is a properly pleaded, well-supported claim for fraud in the inducement, if the customer believes the vendor made misrepresentations to induce the contract and has the evidence to support it. Fraud is not just another count to pile onto the complaint. In the right posture it does two things at once: it opens the door to tort and out-of-pocket damages that the contract tried to foreclose, and — critically — it can render the limitation-of-liability clause itself unenforceable as to the fraudulent conduct. This article explains the mechanics, with California as the primary focus and New York and Massachusetts as the two other jurisdictions where enterprise software disputes frequently end up. Why the cap is the battlefield Oracle’s and SAP’s enterprise agreements are drafted by very good lawyers to do one thing above all: convert a potentially catastrophic delivery failure into a bounded, budgetable liability. The tools are familiar. Oracle’s cloud contracts — the Oracle Master Agreement and the Fusion Cloud Service order documents — pair a damages cap, typically pegged to fees paid over some trailing period, with a sweeping waiver of indirect, special, incidental, and consequential damages, and a disclaimer of lost profits. They are frequently accompanied by an integration or merger clause and, in financed deals, a payment agreement assigned to a third-party lender through Oracle Credit Corporation. SAP’s agreements follow a similar architecture around its software license and cloud/RISE subscription terms. The result is a structural mismatch. A customer’s real-world loss from a failed enterprise implementation is dominated by exactly the categories the contract waives — consequential damages, lost productivity, and the cost of a replacement system. The cap is engineered so that the customer’s provable “direct” damages are a fraction of its true loss. Litigating the breach alone, in other words, is often a losing economic proposition even when liability is clear. That is precisely why the analysis has to start with the cap, not the breach — and why fraud is the lever, provided the customer has real facts to support the claim. The fraud lever: an independent wrong the contract did not price The intuition behind every fraud-based attack on a liability cap is the same across jurisdictions: parties are presumed to have allocated the ordinary risks of performance in their contract, but no one bargains for the right to be lied to. A limitation clause allocates the risk that the software underperforms. It does not, as a matter of public policy in many commercial jurisdictions, allocate the risk that the vendor knowingly misrepresented the software’s capabilities to win the deal in the first place. The evidentiary heart of these cases is almost always the pre-contract sales cycle: the demos, the “solution overviews,” the RFP responses, the emails assuring the customer that the platform will handle its industry “out of the box,” that go-live will happen on a fixed timeline for a fixed price, that little or no customization is required, and that third-party add-ons will not be needed. When the executed order form then describes only components — and omits the capability that was the whole reason for the purchase — the gap between what was sold and what was signed is where the fraud claim lives. California: the primary jurisdiction California is where a large share of these disputes are filed, as many companies including Oracle are present here, and California law is comparatively favorable to defrauded customers. Three doctrinal pillars matter. 1. The economic loss rule does not bar fraud independent of the breach California’s economic loss rule generally bars tort recovery for purely economic losses between contracting parties. But it does not bar an intentional fraud claim based on conduct independent of the breach. In Robinson Helicopter Co. v. Dana Corp., 34 Cal. 4th 979, (2004), the California Supreme Court allowed fraud-based tort remedies where the defendant’s affirmative misrepresentations were independent of the mere failure to deliver conforming goods. The Court reasoned that no rational party enters a contract expecting to be defrauded, and that allowing the economic loss rule to swallow intentional fraud would reward dishonesty. For years, defendants argued that Robinson Helicopter was narrow and confined to affirmative misrepresentations that exposed the plaintiff to personal-injury-type liability. The Supreme Court closed much of that gap in Rattagan v. Uber Technologies, Inc., 17 Cal. 5th 1, (2024). Rattagan confirms that fraudulent concealment arising from or related to contract performance may also sound in tort where the concealment can be established independently of the parties’ contractual obligations and exposes the plaintiff to a risk beyond what the parties reasonably contemplated at contracting. Read together with Lazar v. Superior Court, 12 Cal. 4th 631, (1996), which recognizes tort remedies for fraudulent inducement, and consistent with Erlich v. Menezes, 21 Cal. 4th 543, 981 P.2d 978, 87 Cal. Rptr. 2d 886 (1999), which insists on an independent tort duty before contract claims become tort claims, Rattagan confirms that both affirmative misrepresentations and knowing concealment in the ERP sales cycle can sound in tort. That matters because tort framing is what unlocks out-of-pocket damages and, in egregious cases, punitive exposure that the contract’s damages waiver may not reach. The important limit, reaffirmed in Sheen v. Wells Fargo Bank, N.A., 12 Cal. 5th 905, (2022), is that the tort must be genuinely independent of the broken promise. A repackaged breach — “they promised the software would work and it didn’t” — will not clear the bar. The claim has to rest on a knowing misrepresentation of then-existing fact or a concealed material fact, pleaded with the specificity California fraud pleading demands. 2. Civil Code section 1668 can void the cap itself — even for sophisticated parties This is the development that has changed the leverage in these cases. Cal. Civ. Code § 1668 provides that all contracts that have as their object, “directly or indirectly,” the exemption of anyone from responsibility for their own fraud or willful injury are against the policy of the law. For decades, courts disagreed about whether section 1668 voided only total releases or also reached clauses that merely limited liability. In New England Country Foods, LLC v. Vanlaw Food Products, Inc., 17 Cal. 5th 618, (2025), answering a question certified by the Ninth Circuit, the California Supreme Court resolved the split: Section 1668 invalidates contractual provisions that substantially limit damages for willful injury to person or property, even where the clause is framed as a limitation of liability rather than a complete release. The Court rejected a broad sophisticated-party exception, while preserving ordinary contractual limitations for non-willful breach-of-contract claims. Earlier authority, including Food Safety Net Services v. Eco Safe Systems USA, Inc., 209 Cal. App. 4th 1118, (2012), points in the same direction for fraud-based claims. The practical significance is hard to overstate, but it should be stated precisely. Where the customer can plead and ultimately prove intentional fraud or willful injury within section 1668’s scope — as opposed to a garden-variety breach — the vendor’s fees-paid cap and consequential-damages waiver cannot be used to substantially exempt the vendor from responsibility for that conduct. New England Country Foods was careful to preserve the enforceability of caps for ordinary breach of contract; the statute reaches intentional wrongs within its ambit, not ordinary contract disappointment. So the fraud claim is not just a route to bigger damages. It is the mechanism that can dissolve the cap as to the fraudulent or willful conduct. 3. The integration clause is not the wall the vendor thinks it is Vendors routinely argue that the contract’s merger or integration clause bars any reliance on pre-contract sales representations. In California, the fraud exception to the parol evidence rule answers that argument. Riverisland Cold Storage, Inc. v. Fresno-Madera Production Credit Ass’n, 55 Cal. 4th (2013), overruled the old Pendergrass limitation and confirmed that a party may introduce extrinsic evidence of fraudulent representations to establish fraud in the inducement, even where those representations contradict the written agreement. An integration clause does not immunize a vendor from proof of the promises its salesforce actually made. How this looks in real Oracle litigationThese are not abstractions. A wave of California suits and filings has tested variations of this theory against Oracle’s ERP, HCM, and cloud products. The recurring allegations are familiar: the customer contends that Oracle or its implementation partner misrepresented product capabilities, downplayed the customization required, overstated the partner’s expertise, or promised functionality that did not appear in the executed order documents. In that posture, the breach claim may be vulnerable if the final contract does not actually require the configured, workable solution the customer thought it was buying. The fraud claim, by contrast, may remain viable if the customer can identify specific pre-contract misrepresentations of then-existing fact, the speakers, the timing, reliance, and resulting damage. Recent Oracle and NetSuite disputes illustrate the pattern at the pleading level. Customers have alleged that sales representatives made specific pre-contract promises that a system was tailored to the customer’s industry, would require little or no third-party functionality, or would support critical business requirements. Those allegations are typically paired with breach, fraudulent misrepresentation, negligent misrepresentation, and, in California cases, claims under California’s Unfair Competition Law, Cal. Bus. & Prof. Code § 17200. The lesson is not that every failed implementation becomes fraud. It is that the decisive facts often live in the sales record, not the project plan. The exposure is not limited to NetSuite. Public reporting around failed large-scale Oracle Fusion and SAP programs shows how an enterprise implementation can fail at scale, leaving the customer with impaired operations, a costly remediation project, and losses that ordinary contract caps are designed to exclude. Those examples are useful business context, but they should be distinguished from legal authority. The legal question remains whether the customer can plead and prove fraud or willful misconduct independent of the contract breach. One California-specific trap should be flagged at the outset. Many Oracle deals are financed through Oracle Credit Corporation, with the payment agreement then assigned to a third-party bank. Those financing agreements may contain waiver-of-defenses language governed by UCC Article 9, including Cal. Com. Code § 9403, which can materially affect the customer’s ability to assert software-related defenses against an assignee. The fraud strategy therefore has to account for the financing structure from day one, not after the bank seeks payment. New York: the sophisticated-party jurisdiction New York is the other dominant forum for enterprise software disputes, and its law is meaningfully less forgiving than California’s. The starting presumption favors enforcement. In Metropolitan Life Insurance Co. v. Noble Lowndes International, Inc., 84 N.Y.2d 430, 643 N.E.2d 504, 618 N.Y.S.2d 882 (1994) — itself a software development and installation dispute — the Court of Appeals enforced a negotiated limitation of liability where the vendor’s deliberate, economically motivated refusal to perform was not enough to defeat the cap absent proof of fraud, malice, or other truly willful tortious conduct. New York courts generally hold sophisticated parties to their negotiated risk allocation. The public-policy exception is real but demanding. Under Kalisch-Jarcho, Inc. v. City of New York, 58 N.Y.2d 377, 461 N.Y.S.2d 746 (1983), and related authorities including Sommer v. Federal Signal Corp., 79 N.Y.2d 540, 593 N.E.2d 1365, 583 N.Y.S.2d 957 (1992), a limitation or exculpatory clause generally will not shield conduct that is grossly negligent, willful, or in bad faith. But Colnaghi, U.S.A., Ltd. v. Jewelers Protection Services, Ltd., 81 N.Y.2d 821, 611 N.E.2d 282, 595 N.Y.S.2d 381 (1993), illustrates how high that bar is: gross negligence requires conduct that evinces reckless disregard for the rights of others or smacks of intentional wrongdoing. Fraud in the inducement, well pleaded, is the archetypal conduct that may clear this bar. Ordinary breach, and even deliberate nonperformance for economic reasons, usually does not. The distinctive New York battleground is the anti-reliance clause. Under Danann Realty Corp. v. Harris, 5 N.Y.2d 317, 157 N.E.2d 597, 184 N.Y.S.2d 599 (1959), a specific disclaimer of reliance — a provision in which the buyer represents that it did not rely on any representations outside the four corners of the agreement as to the very matter later alleged to be misrepresented — can defeat a fraudulent inducement claim. A general merger clause, by contrast, does not. The line between the two frequently decides these cases at the motion-to-dismiss stage, which is why the precise wording of the vendor’s “no reliance” language must be scrutinized before filing. There is a crucial but fact-specific escape hatch. Even a specific disclaimer may not bar a fraud claim where the misrepresented facts were peculiarly within the defendant’s knowledge and not discoverable through ordinary diligence. Basis Yield Alpha Fund (Master) v. Goldman Sachs Group, Inc., 115 A.D.3d 128, 980 N.Y.S.2d 21 (1st Dep’t 2014), is useful on that point. In the ERP context, internal defect logs, failed reference implementations, and internal engineering assessments may be candidates for peculiar-knowledge treatment, but the issue will turn on the precise disclaimer language and what diligence was realistically available to the customer. And throughout, New York’s heightened pleading standard, N.Y. C.P.L.R. 3016(b), means the fraud must be alleged with particularity: the specific statements, speakers, and circumstances, not conclusory labels. Massachusetts: Chapter 93A as a damages multiplier Massachusetts deserves its own analysis because it offers a statutory route that can be even more powerful than common-law fraud. Mass. Gen. Laws ch. 93A, § 11 prohibits unfair or deceptive acts in trade or commerce between businesses and authorizes recovery of double or treble damages plus attorneys’ fees for willful or knowing violations. Deceptive pre-contract representations of the kind at issue in ERP disputes are classic Chapter 93A conduct. Decisively, in H1 Lincoln, Inc. v. South Washington Street, LLC, 489 Mass. 1, 179 N.E.3d 545 (2022), the Supreme Judicial Court held that a contractual limitation-of-liability provision will not be enforced to protect a defendant who willfully or knowingly engages in unfair or deceptive conduct prohibited by Chapter 93A. The Court reasoned that the statute’s punitive and deterrent purposes cannot be overridden by private risk allocation, even among sophisticated parties, and even where the clause purports to waive consequential damages — the very category into which Chapter 93A multiple damages fall. The Court declined to let the older, more permissive analysis of Canal Electric Co. v. Westinghouse Electric Corp., 406 Mass. 369, 548 N.E.2d 182 (1990), shield willful violators, and distinguished the tort/contract analysis in Standard Register Co. v. Bolton-Emerson, Inc., 38 Mass. App. Ct. 545, 649 N.E.2d 791 (1995). Massachusetts also has directly relevant software authority. In VMark Software, Inc. v. EMC Corp., 37 Mass. App. Ct. 610, 642 N.E.2d 587 (1994), the Appeals Court affirmed misrepresentation-based relief arising from assurances about software functionality, while declining multiple damages on the facts. The lesson is familiar: the quality of the misconduct — merely misleading versus willful or knowing — drives both whether contractual limitations yield and whether damages multiply. Two practical cautions. First, a section 11 claim requires that the unfair or deceptive conduct occurred “primarily and substantially” within Massachusetts; courts examine the center of gravity of the misconduct, and an out-of-state vendor will press this hard. Second, a defendant can blunt the multiplier by making a reasonable single-damages settlement offer with its answer, so the timing and content of pre-suit Chapter 93A demand letters and early settlement posture genuinely affect the exposure. Practical takeaways for enterprise customers and their counsel The through-line across all three jurisdictions is that the fraud claim, not the breach claim, is what can make a failed enterprise implementation economically worth litigating. To preserve it:
Celonis v. SAP Update: What Celonis's Proposed Second Amended Complaint Adds to the Lawsuit5/11/2026 By Pam Fulmer
On May 2, 2026, Celonis filed a motion in Celonis SE v. SAP SE, Case No. 3:25-cv-02519-VC (N.D. Cal.), for leave to file a Second Amended Complaint. The proposed pleading — running 117 pages — has not yet been approved by Judge Vince Chhabria, but it is publicly filed and worth reading. Stripped of the antitrust framing, the proposed complaint is a detailed account of an enterprise software vendor squeezing its installed base. Here is what is alleged, organized by what is happening to the customer rather than by claim. The Customer Owns the Data, But Cannot Get to It Celonis alleges that SAP's own General Terms and Conditions for Cloud Services confirm that customers own their enterprise data. The data resides in the customer's instance of the SAP ERP system, often on the customer's own servers. Celonis's process-mining tool reads that data from the customer's environment. SAP is not in the path. Notwithstanding that ownership structure, the proposed complaint walks through a sequence of SAP technical notes — each of which incrementally narrowed how a customer could extract its own data for use with a non-SAP tool. Celonis alleges that after the sequence of changes, only two extraction pathways remain technically permitted: OData, which SAP has withdrawn support for, and Datasphere, an SAP product that the complaint alleges carries fees so high that using it as an extraction conduit to a non-SAP tool is commercially infeasible — sometimes exceeding the cost of the third-party tool itself. The customer-side picture this paints is a familiar one. A customer that purchased SAP ERP under a set of expectations about its ability to work with the third-party tools of its choice has watched those expectations narrow through a succession of vendor-published technical notes the customer never specifically agreed to. The customer's contract did not change. The technical implementation of the customer's contract did. SAP Allegedly Lied to its Customers to Get Them to Stop Using a Vendor They Liked The most operationally consequential allegations in the proposed amended complaint are the false-statement allegations. Celonis identifies — by individual SAP employee, by date, by customer, and by substance — specific communications in which SAP told joint customers that using Celonis would violate license terms, require purchase of additional database or HANA full-use licenses, render the customer non-compliant with SAP policy, or imperil S/4HANA migration. The proposed pleading references internal SAP materials that, according to Celonis, instructed SAP account teams to make these statements on a "case-by-case basis" and to avoid any "general compliance campaign" or "public communications" — a directive Celonis cites as evidence that SAP itself recognized the statements were problematic. For the customers on the receiving end of those conversations, the experience was not abstract. The proposed complaint reflects customer inquiries to Celonis in which the customer reported being told its current extraction processes were "no longer permitted," asked Celonis whether it was "in compliance with SAP requirements," reported having been "warned" about future problems with Celonis after migrating to S/4HANA, and asked whether SAP would "want to charge us" for the data connection. One customer told Celonis that as a result of what SAP had communicated, the customer "may be unable to implement Celonis" on its S/4HANA instance. Another reduced its Celonis contract because of its understanding that the Celonis approach "is not allowed." These are not antitrust harms in the abstract. They are operational decisions enterprise customers made based on information the proposed complaint alleges was false. Customers Were Pushed Into a Bundle Whether They Wanted It Or Not Celonis alleges that SAP has been giving Signavio — its own process-mining product — away free or near-free inside the RISE bundle, with an internal directive to "include Signavio in every software sale." The customer effect is that customers renewing or expanding their SAP relationship receive Signavio at no additional incremental cost, and Celonis alleges that this has caused customers to drop or scale back their Celonis usage in favor of a product the proposed complaint characterizes as inferior. Celonis identifies multiple lost expansion and renewal contracts — customer names redacted — where Celonis was told the reason for the loss was Signavio's inclusion in a RISE bundle. For customers, the dynamic is one we see repeatedly in enterprise software. A bundled component is presented as free, which makes it operationally rational to consume it. The free component then displaces a third-party tool the customer had been paying for. The customer "saves money" in the short run and loses optionality in the long run, as the third-party market shrinks and the vendor's bundled offering becomes the default. Customers Are Migration Hostages Threaded through the proposed pleading is an allegation that resonates with our practice and bears separate attention: customers' relationships with third-party tools are being disrupted at the exact moment customers are migrating to S/4HANA, and SAP is using the migration as leverage. The proposed amended complaint alleges that SAP communicated to customers that continued use of Celonis could jeopardize their S/4HANA migration — a migration most enterprise SAP customers cannot realistically defer. The proposed pleading frames this as part of a coercive scheme. The customer-side characterization is simpler: the customer cannot exit, cannot defer the migration, and is making procurement decisions about third-party tools under conditions in which the dominant vendor is communicating that the customer's strategic IT future depends on cooperating. That is the textbook fact pattern of economic duress under Rich & Whillock, Inc. v. Ashton Development, Inc., 157 Cal. App. 3d 1154 (1984), and an important reason customers facing this type of vendor pressure should preserve communications carefully and consider counsel involvement early. Customer Choice — Once Promised, Now Allegedly Removed The proposed complaint quotes SAP's own historical promises of an "open ecosystem" and "free customer choice," made publicly between 2012 and 2018, on which Celonis and other third-party developers built their businesses. The same proposed complaint alleges that SAP continues to advertise its platform as an "open ecosystem" on customer-facing materials, while internally directing the conduct described above. The complaint also notes that SAP gave explicit assurances to antitrust regulators reviewing the Signavio acquisition that process-management software like Signavio would require only "scanner access" with no fees for indirect use applicable — assurances Celonis says SAP has not honored. For customers, the gap between vendor public messaging and vendor account-team conduct is a recurring theme. The proposed complaint illustrates how that gap can be documented and ultimately litigated. What This Means for SAP Customers Right Now A few practical observations follow from reading the proposed Second Amended Complaint as a customer-side document. First, document everything. The proposed pleading is built on emails, sales-team communications, internal SAP slides, and customer-to-vendor inquiries. The customer that preserves these communications — whether or not it ever intends to litigate — is the customer with the strongest hand at the next renewal. Second, treat compliance assertions skeptically. The proposed complaint alleges that the central pattern of SAP's customer-facing campaign was false statements that the customer's use of Celonis was non-compliant, would trigger additional license requirements, or would create technical or migration risk. Customers who hear similar assertions from any enterprise vendor should obtain those assertions in writing and route them through counsel before acting on them. Vendor compliance claims are not self-validating. Third, recognize what California law provides. We have written before about the California-law tools available to customers facing aggressive vendor conduct: the implied covenant of good faith and fair dealing under Carma Developers (Cal.), Inc. v. Marathon Development California, Inc., 2 Cal. 4th 342 (1992); California's Unfair Competition Law under Business and Professions Code section 17200; economic duress under Rich & Whillock; and tortious interference theories where the vendor's conduct disrupts the customer's relationships with third parties. The Celonis litigation is a live test of how these tools apply to enterprise software conduct. The legal infrastructure California customers can use is more developed than is often appreciated. Fourth, the third-party tools customers depend on may have claims of their own. Celonis is litigating in its own name, but the conduct it describes is conduct directed at SAP customers. Customers whose preferred third-party tools have been the target of similar vendor pressure should be aware that the third party may have independent claims, and that the customer's documentation may be relevant evidence in those proceedings. Caveats Two important ones. First, this is a proposed pleading. The Court has not yet granted leave to amend; until it does, the First Amended Complaint remains the operative pleading. Second, these are allegations only. SAP has denied the allegations in its responsive pleadings and will be entitled to test them through discovery, dispositive motions, and ultimately at the December 7, 2026 trial. Nothing in this post should be read as a finding or conclusion about the merits. Closing Thought The most important sentence in the proposed amended complaint, from the customer's perspective, may be the one in which Celonis frames its own theory: customers buy SAP's ERP software to collect and run their own data, and SAP is allegedly using its control over that ecosystem to deny customers the freedom to work with the providers of their choice. Whether or not Celonis ultimately proves its case, the underlying dynamic — a dominant enterprise vendor narrowing customer choice through technical, contractual, and informational levers — is one California licensees will continue to encounter. The proposed pleading is a useful map of what that dynamic looks like in practice and where the legal pressure points are. We will continue to monitor the case as the Court rules on the motion to amend. Tactical Law Group LLP represents enterprise software licensees in licensing disputes, audit defense, and commercial negotiations involving Oracle, SAP, Broadcom, and other enterprise software vendors. Nothing in this post is legal advice or a comment on any pending litigation. The allegations described are taken from a publicly filed proposed pleading and have not been adjudicated. If your organization is facing aggressive vendor conduct directed at your relationships with third-party providers, please contact us directly. By Pam Fulmer
Software audit disputes used to be uncomfortable but survivable. A publisher would audit usage, claim over-deployment, and demand a true-up. The customer could dispute the findings, involve counsel, and negotiate while the business kept running. That leverage has changed. In a SaaS, cloud-hosted, subscription, or remotely administered environment, the vendor may control the customer’s practical ability to operate. If the vendor can suspend access, disable authentication, block a hosted environment, or lock down critical data, the dispute is no longer just about who is right under the contract. It is about whether the customer can keep running long enough to find out. California law has not yet developed a mature body of published SaaS “kill switch” cases. But California does provide a set of doctrines that can matter when a vendor threatens to disable mission-critical software to collect a disputed demand: the implied covenant of good faith and fair dealing, economic duress, unconscionability, conversion and trespass to chattels in appropriate cases, the Unfair Competition Law, and emergency injunctive relief. The key is precision. The customer’s argument should not be that a vendor can never suspend service. If the contract clearly allows suspension after defined conditions are met, California courts will generally take that language seriously. The stronger argument is that a vendor may not use a suspension right beyond its contractual scope, in bad faith, without satisfying conditions precedent, to enforce a knowingly inflated demand, or in a way that interferes with customer-owned property or data beyond what the agreement permits. And as a lawyer defending software audit disputes for years, I can tell you that many demands are knowingly and intentionally inflated to use as leverage to extract a large software purchase from the customer. Now imagine how empowered these same predatory publishers will be with the kill switch in their hands. Enterprise software customers would do well to start planning their strategies now. Start with the Contract The first question is what the agreement actually says. Many enterprise software agreements contain suspension provisions for nonpayment, uncured breach, security risk, license overuse, audit noncompliance, or violation of acceptable-use restrictions. Some clauses are narrow and procedural. Others are broad and vendor-friendly. California law gives real force to express contract language. In Carma Developers (Cal.), Inc. v. Marathon Development California, Inc., 2 Cal. 4th 342 (1992), the California Supreme Court held that the implied covenant of good faith and fair dealing may not be used to prohibit conduct the agreement expressly permits. The covenant protects the bargain; it does not rewrite it. In Bevis v. Terrace View Partners, LP, 33 Cal. App. 5th 230 (2019), the Court of Appeal likewise rejected an implied-covenant theory that would have required a party to choose one contractually permitted course over another. Those cases are important vendor-side authority. But Carma also identifies the customer’s opening: the implied covenant has particular force where one party holds discretionary power affecting the rights of the other. The strongest customer argument is not “the vendor can never suspend.” It is that the vendor cannot use a discretionary suspension mechanism as a pretextual coercion device to obtain benefits outside the bargain or to enforce a claim it knows is false or materially overstated. Think of Oracle's VMware virtualization policy arguments. Relevant questions include whether the contract allowed suspension for this type of alleged breach; whether the vendor satisfied notice, cure, audit, escalation, and dispute-resolution requirements; whether the customer is current on undisputed amounts; and whether the vendor is threatening to suspend services, data, affiliates, or environments beyond the clause’s scope. Economic Duress California’s economic-duress doctrine can be powerful in a kill-switch dispute, but only if the customer can show more than ordinary commercial pressure. The leading case is Rich & Whillock, Inc. v. Ashton Development, Inc., 157 Cal. App. 3d 1154 (1984), where a release was unenforceable because it was obtained after a contractor refused to pay an undisputed amount, knowing the other party faced financial ruin. The doctrine turns on a wrongful act, coercive pressure leaving no reasonable alternative, and submission to the pressure. The wrongful act need not be a crime or independent tort; asserting a claim known to be false, making a bad-faith threat to breach, or wrongfully withholding payment may qualify. That framework can fit software suspension threats where a vendor uses a knowingly inflated audit claim, refuses to follow agreed procedures, threatens suspension for amounts not yet due, or demands unrelated purchases or broad releases as the price of continued access. But a vendor’s threat to exercise a clear contractual suspension right after a material uncured breach is not automatically wrongful merely because it creates pressure. If payment is necessary to keep operating, the customer should pay expressly under protest, reserve all rights, identify the disputed grounds in writing, and document why no reasonable alternative existed. Unconscionability Unconscionability is possible, but often difficult in enterprise software disputes. In Sanchez v. Valencia Holding Co., 61 Cal. 4th 899 (2015), the California Supreme Court explained that unconscionability requires both procedural and substantive elements. Sonic-Calabasas A, Inc. v. Moreno, 57 Cal. 4th 1109 (2013) describes the same basic framework. In B2B software contracts, the customer may be sophisticated, represented, and able to evaluate alternatives. The better argument targets the combined remedial architecture: unilateral vendor breach determinations, short cure periods, suspension before neutral review, a bar on consequential damages, a low liability cap, and no meaningful data-export right. The goal may be limited but important: preserve access during a good-faith dispute or prevent the vendor from invoking a liability cap for a wrongful shutdown. Conversion, Trespass, and Data Lockout Tort theories are strongest when the vendor interferes with property interests beyond a mere contractual right to use the vendor’s hosted service. The customer should identify exactly what property is being impaired: customer data, electronic records, backups, local installations, servers, devices, credentials, or domain assets. In Intel Corp. v. Hamidi, 30 Cal. 4th 1342 (2003), the California Supreme Court held that unwanted emails did not establish trespass to chattels because they did not damage Intel’s computer system or impair its functioning. For kill-switch purposes, Intel underscores the need for concrete impairment: blocked access, disabled functionality, interruption of system use, or loss of access to records. Conversion may also apply to certain digital property. In Kremen v. Cohen, 337 F.3d 1024 (9th Cir. 2007), the Ninth Circuit, applying California law, held that a domain name could support a conversion claim. The out-of-state case Clayton X-Ray Co. v. Professional Systems Corp., 812 S.W.2d 565 (Mo. Ct. App. 1991) remains useful by analogy, but California briefing should lead with California authority. UCL and Emergency Relief California’s Unfair Competition Law can support restitution and injunctive relief, but not ordinary damages. Korea Supply Co. v. Lockheed Martin Corp., 29 Cal. 4th 1134 (2003) makes that remedial limit clear. The “unlawful” prong is often the cleanest route, using breach of the implied covenant, duress, unconscionability, statutory violations, or wrongful property interference as predicates. The “unfair” prong requires caution in B2B disputes; under Cel-Tech Communications, Inc. v. Los Angeles Cellular Telephone Co., 20 Cal. 4th 163 (1999), unfairness in competitor cases must be tethered to a legislatively declared policy or threaten competition. The most important remedy may be a temporary restraining order or preliminary injunction. California courts consider likelihood of success and the interim harm to each side. Butt v. State of California, 4 Cal. 4th 668 (1992). The customer should build a record showing both irreparable harm and merits: disputed demand, payment of undisputed amounts, failure to follow contract procedures, operational dependency, lack of alternatives, and threatened loss of customer-owned data. California law gives customers a toolkit, not a silver bullet. The winning case usually turns on contract language, procedural defects, a timely good-faith dispute, wrongful pressure, and whether suspension would interfere with customer data or operations in a way damages cannot repair. That is enough to change the negotiation. A vendor facing a prepared TRO application, a duress record, possible UCL restitution, and property-based claims for overreach must evaluate the cost of flipping the switch more carefully. This article is for general informational purposes only and does not constitute legal advice. Readers should consult counsel about the specific facts of their own situation By Pam Fulmer
A finance manager at a California company opens Microsoft 365 Copilot and asks, “Summarize our open purchase orders over $50,000 and flag anything unusual.” Copilot reaches SAP through a connector or other integration layer, builds the answer, and drops it into a dashboard she shares with four hundred colleagues. She is a licensed SAP user. The four hundred colleagues are not. Is that indirect access? Does each of those four hundred colleagues now need an SAP Named User license? Does it matter if Copilot creates a purchase order on her behalf rather than just summarizing one? What if the whole workflow runs autonomously, with no human prompting the agent at all? There is no clear AI-specific published guidance that answers those questions — and that is precisely the problem. SAP and Oracle are actively marketing AI-enabled products and integrations, customers are deploying them at speed, and the contract language that will ultimately be used to assess licensing exposure was drafted decades before anyone imagined an AI agent as a user. This Is Not a New Problem — It Is the Diageo Problem in New Clothes In 2017, the UK High Court ruled in SAP UK Ltd v. Diageo Great Britain Ltd that thousands of Diageo customers and sales representatives using Salesforce-based apps — apps that in turn exchanged data with SAP — constituted indirect users of SAP ERP. SAP claimed more than £54.5 million in additional license fees. The case settled before the damages phase, but the liability ruling became a touchstone for later indirect-access disputes and helped catalyze SAP’s move toward its current Digital Access model, while reinforcing the broader vendor view reflected in Oracle’s aggressive enforcement of its multiplexing rule. Diageo was ultimately a case about middleware. The court concluded that even though Salesforce users never logged into SAP directly, the fact that their actions flowed through middleware into SAP meant they were using the SAP software. That reasoning — that “use” and “access” can reach through whatever sits in the middle — is exactly what makes AI agents a likely next battleground. What’s Different Now: The AI Agent Fact Patterns AI agents raise the indirect access problem in four distinct ways, and each one introduces contractual ambiguity the vendors have not resolved. First, conversational assistants with ERP connectors. Microsoft 365 Copilot, ChatGPT with custom connectors, and similar tools allow a licensed user to query SAP or Oracle data and then redistribute the result to an unlimited audience. The licensed user pays for the license, but the practical beneficiaries may be hundreds of colleagues who never had a seat. Second, agentic workflows that create transactions autonomously. Procurement-to-pay pipelines can match invoices to purchase orders and post documents into S/4HANA without a human in the loop. Under SAP’s Digital Access model, every posted document — sales order, invoice, purchase order, journal entry, and others — may be a countable event. An agentic pipeline can multiply a customer’s historical document volume many times over, and SAP’s published materials do not clearly exclude AI-generated documents from the count. Drafts, retries, and reversal entries only compound the issue. Third, service-account connections in Oracle environments. A customer-service AI agent might use a single Oracle service account to answer billing or shipment questions on behalf of thousands of end customers. Oracle’s multiplexing rule, essentially unchanged for years, states that multiplexing does not reduce Oracle license requirements and that users at the multiplexing front end must still be licensed. On its face, that rule could be read to reach every one of those end customers — a potentially devastating position in an audit. Fourth, retrieval-augmented generation pipelines. A common enterprise pattern now is to extract master data from SAP or Oracle nightly, embed it into a vector database, and answer employee questions from the vector store. Is the nightly extract the relevant “access” event — a single licensed pathway? Or does every downstream question count because the data originated in SAP or Oracle? The contract language usually does not resolve that issue, and a motivated vendor auditor can argue it either way. The Vendor Silence Is Deliberate SAP now offers Joule base capabilities at no additional cost, while pricing certain premium AI capabilities separately, including in some cases through consumption-based AI Units. Oracle has embedded hundreds of AI agents across its Fusion Cloud applications. Both vendors are actively marketing these capabilities. Neither vendor, however, has published clear guidance answering the licensing questions above. That silence is a feature, not a bug. Ambiguous contract language is one of the most powerful tools a licensing team has in an audit. When the rules are unclear, the vendor gets to assert the most expensive reading first and negotiate downward from there. Customers that did not think to negotiate AI-specific language in their 2019 or 2022 renewals are the ones most exposed. Why California Customers Have Leverage California is home to a disproportionate share of enterprise SAP and Oracle customers, and California law gives customers several tools when a vendor tries to stretch pre-AI contract language to cover an AI deployment the parties never discussed. Every California contract carries an implied covenant of good faith and fair dealing. Where one party holds discretionary power — as vendors often do when interpreting their own license terms — that discretion must be exercised reasonably and with proper motives. A vendor that assesses a multi-million-dollar compliance finding against a customer on a theory the parties never discussed at signing may face a substantial good-faith challenge. California Civil Code section 1654 codifies the rule that ambiguous contract language is construed against the drafter. California courts apply that principle seriously, particularly in agreements drafted by sophisticated legal teams — and SAP and Oracle agreements are drafted by some of the most sophisticated licensing teams in the industry. Ambiguous words like “user,” “access,” or “multiplexing front end” belong to the vendor. If the vendor intended those terms to cover AI agents, copilots, downstream recipients, or vector-database architectures, it should have said so in the contract rather than for the first time in an audit demand. California’s Unfair Competition Law, Business and Professions Code section 17200, reaches unlawful, unfair, and fraudulent business practices. It can be especially useful when a vendor changes its interpretation of the same contract language between customers or over time, or when audit conduct crosses the line into misrepresentation or concealment. Finally, course of performance matters. If a vendor audited a customer in 2022 and did not flag an AI-enabled integration that was already in place, that audit history may support the customer’s interpretation of the contract and may strengthen waiver, estoppel, or course-of-performance arguments when the same vendor audits the same integration in 2026 and suddenly claims a compliance failure. Customers should be preserving audit history, support tickets, and account-team correspondence now, while memories are fresh, rather than scrambling later. Getting Ahead of the Problem There are a handful of practical steps every SAP or Oracle customer deploying AI agents should take before the first audit letter arrives. Revisit your most recent vendor contract and study the definitions of “user,” “access,” “indirect use,” and — for Oracle — “multiplexing.” Where the language is silent on AI agents, that silence is both an argument for you in the short term and a redline target in the next renewal. If the vendor is offering a new AI product as an add-on, insist on written confirmation of how that product interacts with your existing license metrics before you buy. Document your AI deployment architecture now. How does the agent connect? Who are the prompting users? What does the agent read, and what does it write? Which outputs create countable documents under Digital Access, and which are merely transient summaries? The time to build that file is before a vendor audit team builds it for you. Treat vendor-native AI differently from third-party AI. SAP’s Joule and Oracle’s Fusion AI agents are the vendor’s own products. There is a strong argument that licensing a vendor’s AI features should come with bundled indirect-access rights for the downstream outputs those agents produce. That argument should be made in writing, and it should appear in the contract itself. How Tactical Law Can Help The intersection of AI deployment and ERP licensing is where two fast-moving areas collide, and the customers who will pay the least are the ones who start the conversation before the vendor does. Tactical Law advises enterprise customers on licensing strategy, audit defense, and contract negotiation across exactly these issues. If your organization is deploying AI agents, copilots, or agentic workflows against an SAP or Oracle estate — and you have not yet had a conversation with outside counsel about what that means for your license position — now is the time. Oracle’s Newest Java Audit Demand: Your VMware Topology — and What California Law Says About It4/19/2026 By Pam Fulmer
A pattern is appearing in Oracle’s Java licensing enforcement that every in-house counsel with an Oracle footprint needs to understand. On the sales side, at least in some instances, Oracle is offering customers what is, in substance, a two-track choice. Customers willing to subscribe on the new per-employee Java SE Universal Subscription metric can do so without producing information about their virtualization environment. Customers who want to remain on — or return to — Oracle’s legacy Named User Plus or Processor-based Java metrics may be required by Oracle to first disclose extensive data covering the entire VMware farm, not only the servers where Oracle software is installed or actually running. A Java licensing conversation is, in other words, being converted into a VMware full environment disclosure. The scope of that demand is the tell. Even under the legacy Named User Plus and Processor options, Java compliance is verified by reference to the servers where Oracle Java is actually installed and/or running. When Oracle asks for data about the full virtualized environment — including hosts that do not run Oracle software at all — the data is being collected for a different purpose. This post explains that purpose, why it is dangerous, and the California legal arguments customers can use to push back. What Oracle Is Asking For The audit-side of this pattern is now documented in the trade press. Redress Compliance has reported that Java audit letters ask for “a full list of all VMware or other virtualized platform hosts, whether they have Java installed or not”. House of Brick has documented Oracle asking for vCenter exports and cluster configuration data during Java audits and tying those requests back to Oracle’s aggressive position on VMware licensing. And The Register’s 2024 coverage of Java audit letters to Fortune 100 companies signaled the scale of the escalation. The structure of the choice Oracle is offering customers with meaningful Java dependencies deserves a closer look, because it functions as a Hobson’s choice. Accepting the per-employee metric avoids any VMware inquiry, but it has made Oracle Java dramatically more expensive for most enterprises than the legacy arrangements. Declining that metric in favor of Named User Plus or Processor-based licensing may require the customer to hand over data on the full VMware environment — including hosts that have nothing to do with Oracle software. And walking away from Oracle Java altogether is, for many customers, not a short-term option: a disciplined migration to OpenJDK or another supported distribution takes time, requires engineering and testing work, and introduces business risk that cannot be absorbed on Oracle’s negotiation timeline. Customers have understandably balked at the VMware-disclosure path. Producing whole-farm topology to Oracle at any stage of a Java engagement raises the risk that the inquiry will expand beyond Java, or that Oracle will use the data to assert compliance claims about other Oracle products running in the same environment — most obviously Oracle Database. That is the subscription-side extension of the pattern we described in “Oracle Java Licensing Enforcement: How ‘Friendly Outreach’ Is Driving Significant Compliance Risk” and in “How Oracle Uses Online Agreements for ‘Free Software’ to Trap Companies”: Oracle’s outreach is not just pre-litigation intake — in some instances it has become pre-audit intake, with the subscription transaction itself used as the lever. Why It Is Dangerous The purpose of the VMware request is Oracle’s long-running “soft partitioning” position on database licensing — the whitepaper theory, never codified in customer agreements, that any physical core in a VMware cluster where Oracle software could theoretically run must be fully licensed. Under its more aggressive expressions, according to Oracle, every host connected to the same vCenter, or reachable by vMotion, must be licensed for any Oracle software running anywhere in the environment. For a customer running a modest Oracle Database footprint on a large VMware estate, the resulting compliance gap is often very large. That position has never been tested in court with a court ruling, and independent specialists have argued forcefully that Oracle’s soft-partitioning theory is inconsistent with how VMware actually works. But the economic pressure to settle rather than litigate is enormous, and Oracle knows it. A customer who hands over complete vCenter topology during a Java audit has, in practical terms, already pre-calculated the database compliance claim Oracle will assert three months later. The Java audit is the delivery vehicle. The database claim is the payload. California Legal Arguments That Matter For Oracle customers — many of whom operate under Oracle agreements that select California law by an express choice-of-law provision — California provides a toolkit for pushing back on this conduct. As California lawyers, we are intimately familiar with this toolkit. The Unfair Competition Law, Business & Professions Code § 17200, is the most flexible and most important of those tools. Section 17200 prohibits any “unlawful, unfair, or fraudulent business act or practice.” The “unfair” prong reaches conduct that violates public policy or causes substantial injury, even where no specific statute has been violated. Conditioning the sale of a Java subscription — priced on a metric entirely unrelated to virtualization — on the customer’s disclosure of VMware topology that will predictably be used to construct a separate, much larger claim appears to fit the “unfair” framework cleanly. Post-Proposition 64, a UCL plaintiff must show actual injury; a customer who paid an inflated subscription price, or who was forced into a database compliance settlement the disclosure made possible, can satisfy that requirement. The implied covenant of good faith and fair dealing is a second, and often underused, angle. Every California contract includes an implied covenant prohibiting either party from acting to deprive the other of the benefits of the bargain. When Oracle invokes the audit clause from one agreement — an Oracle Master Agreement, a database OLSA, or an OTN license — to extract information whose only function is to build claims under a separate product line, the implied covenant may be available as a basis for a claim. Audit rights exist to verify compliance with the agreement that granted them. Using them as reconnaissance for a different product’s claims is not what the parties agreed to, and California courts take that distinction seriously. Finally, economic duress. California recognizes the doctrine where one party uses a wrongful act or threat to force another into a transaction it would otherwise refuse, and where the coerced party has no reasonable alternative. Rich & Whillock, Inc. v. Ashton Development, Inc. (1984) 157 Cal.App.3d 1154 remains the foundational authority. The choice Oracle is presenting — an expensive new metric, or a whole-VMware farm disclosure that will foreseeably build claims elsewhere, or abandoning a business-critical platform on an infeasible timeline — fits that framework. Most often the scope of Oracle’s demanded disclosure has no legitimate relationship to the Java transaction, and a customer whose Java dependencies cannot be unwound on Oracle’s timeline has no reasonable alternative. Duress is a particularly valuable defense because it attacks the enforceability of any settlement Oracle later extracts from data produced under coercion. What To Do When the Pattern Appears A few practical steps apply whether the demand arrives in a formal audit letter, a GLAS follow-up, or a sales-team email holding up a subscription quote. Stop providing VMware information in any Java communication. Demand in writing that Oracle identify the specific contract clause authorizing the request and the specific Oracle product whose compliance is being verified; if Oracle cannot answer, the request is a fishing expedition. Document any conditioning of a subscription sale on disclosure — that documentation is the foundation of any UCL, implied-covenant, or duress argument later. And involve counsel before information leaves the company. Early, counsel-led responses are the single strongest predictor of a favorable outcome in this pattern. Closing Thought The Java audit is increasingly not about Java. Oracle’s enforcement program is a data-gathering operation with a sales objective attached, and the whole-farm VMware demand is the most aggressive expression of that strategy we have yet seen. California law gives customers real tools to resist it — but those tools only work if the customer reaches for them before the data has been delivered. Tactical Law Group LLP represents enterprise software licensees in Oracle and other software publisher licensing matters, audit defense, and commercial negotiations. Nothing in this post is legal advice or a comment on the specific circumstances of any customer or transaction. If your organization is facing an Oracle Java audit — or is being told a Java subscription is conditioned on VMware or other environmental disclosure — please contact us directly. |
By Tactical Law Attorneys and From Time to Time Their Guests
|
RSS Feed
tactical law group llp |
COPYRIGHT TACTICAL LAW GROUP LLP 2016-2026. ALL RIGHTS RESERVED
|
Legal Disclaimer: Contents may contain attorney advertising under the laws of some states. Prior results do not guarantee a similar outcome.