The world’s most profitable law firm just voted with its checkbook on AI vendor contracts — and it voted to walk away. As Dr. Alex Wissner-Gross reported in The Innermost Loop on May 29, 2026, Kirkland & Ellis has set aside $500 million to build its own AI tools rather than rent them from its rivals. In the same week, Meta started charging for AI features at $7.99 a month. Intelligence is officially a metered utility — and your AI vendor contracts are the meter.
If a firm with effectively unlimited cash is willing to spend nine figures to escape someone else’s AI terms of service, the build-versus-buy question has arrived at every other business too. Here is what that decision actually looks like — and the six contract risks you should review before you sign the next SaaS renewal.
What Kirkland’s $500 Million Decision Actually Signals
Kirkland did not announce a science project. It announced a capital allocation: a half-billion dollars to own the model rather than rent it. The phrase from The Innermost Loop is worth pinning to the wall — “owning the model is the new owning the firm.”
For two years, “AI strategy” mostly meant picking a vendor. The K&E move says the math has flipped. When the price of intelligence is rising, the terms of service are tightening, and the data flowing through the tool is the most valuable thing in the business, leasing the brain becomes a strategic liability. That logic does not stop at law firms. It applies to any company whose product is judgment, design, code, or analysis — which, in 2026, is most of them.
What This Means for Your Business
You are probably not going to spend $500 million on your own foundation model. That is not the lesson. The lesson is that your AI vendor contracts are now strategic documents, not procurement paperwork. They govern three things that determine the future of your company:
First, your costs. AI pricing is moving from flat per-seat to metered usage — tokens consumed, calls made, outputs generated. A clause buried in a renewal can quietly change your unit economics by the next quarter.
Second, your data. Every prompt, every uploaded document, every customer record passed to an AI tool sits inside someone else’s training and retention policy. The default settings usually favor the vendor.
Third, your leverage. As K&E recognized, the firm that depends on a competitor’s AI tool to do its work has handed its competitor a kill switch. The same is true if your dental group, music label, or healthcare practice runs on an AI stack a rival could buy.
The Legal Impact: 6 Risks Hidden in Today’s AI Vendor Contracts
Most AI vendor contracts on the market today were drafted in 2023 or earlier. They have not caught up to how the technology — or the leverage — actually works. Here are the six clauses our corporate practice flags first in any AI vendor agreement review.
1. Data ownership and training rights
Read the data clause carefully. Many vendors retain a license to use your inputs to improve the model, sometimes broadly enough to derive value that competes with you. Look for opt-out language, retention windows, and whether “aggregated” or “anonymized” data is excluded — because in practice it usually is not.
2. Confidentiality and trade secret exposure
If your employees paste client lists, deal terms, or formulas into an AI tool, that disclosure can compromise the confidentiality protections that keep something a trade secret under federal and state law. Your vendor contract should expressly preserve the confidential character of inputs and limit personnel access on the vendor side.
3. Output ownership and IP indemnity
Who owns what the model produces — you, the vendor, or no one? And if the output infringes a third party’s copyright, who pays? IP indemnities in AI contracts vary wildly. Some vendors offer broad protection; others disclaim everything. This is the single most negotiated clause in our AI contract reviews.
4. Pricing change and “metered intelligence” risk
Meta’s $7.99 plan and the K&E story are part of the same shift: AI is being repriced in front of you. Look for caps on price increases, notice periods, and lock-in pricing for any usage you have already committed to. If your contract says “fees may be adjusted upon notice,” you do not have a contract — you have a meter.
5. Termination, portability, and the “kill switch” problem
Can you leave with your data, your fine-tunes, and your prompt library intact? Or is everything you have built inside the vendor’s system stuck there? Termination-assistance clauses, export rights, and transition windows are the difference between a vendor relationship and a hostage situation.
6. Liability allocation and the autonomous-agent question
The next generation of these tools acts on your behalf — sending emails, executing trades, signing documents. When an AI agent takes a costly action, who is liable? Most current vendor contracts allocate that risk entirely to the customer. Our prior analysis of AI agent contract liability walks through how that allocation should actually be negotiated.
What Howard East Clients Should Do Now
You do not need a $500 million budget to act on this. You need a one-hour review of every active AI vendor contract in your business, with the six clauses above in front of you. Specifically:
Pull the AI tools your team actually uses — not the ones procurement signed, but the ones your people log into. Map each to a vendor contract or terms of service. Score each against the six risks. Flag the renewals coming up in the next ninety days. Those are the ones where you still have negotiating leverage.
If your business is in a regulated field — employment, healthcare, financial services, entertainment — bring counsel into the review. Sector-specific rules layer on top of standard SaaS terms, and the ABA’s professional responsibility framework is only one of many regimes now reaching into AI use. For partnership and M&A contexts, AI vendor exposure is also becoming a standard due diligence item — see our guidance on private placement and disclosure obligations and the broader shift in AI in legal services for clients evaluating deals.
If a dispute is already brewing over an AI vendor — pricing change, data leak, output liability — that crosses into litigation, our colleagues at Howard Law Group handle the contested side. For operators in regulated industries thinking through AI alongside compliance build-out, Collateral Base covers the operational consulting layer.
The Bottom Line on AI Vendor Contracts
Kirkland & Ellis spending $500 million to escape AI vendor contracts is not the story. The story is that one of the most cost-conscious institutions on the planet looked at the math and decided that owning beat renting. You do not have to make the same call. You do have to look at the same math — and your AI vendor contracts are where the math lives.
Howard East represents business owners, employers, partnerships, and entertainment clients negotiating and litigating exactly these agreements. If your AI vendor stack has grown faster than your contract review, book a consultation and we will run the six-risk review on every active agreement.
This article is for informational purposes only and does not constitute legal advice. Attorney Advertising.
