Anthropic’s Bold Play: Lock In Bigger AI Discounts—But Only If You Bet Big on Their Tech!

Anthropic’s Bold Play: Lock In Bigger AI Discounts—But Only If You Bet Big on Their Tech!

Ever wonder what it’s like to get a discount on your AI spending… but only if you promise to keep feeding that same vendor a steady diet of your budget? Well, welcome to the high-stakes world of enterprise AI deals, where Anthropic isn’t just asking for your business—they want to slice off a guaranteed portion of your entire AI pie before you can even think about snagging those juicy discounts. It’s a bold play, really, tightening the screws just as Anthropic and OpenAI jostle for the biggest fish in the corporate pond. But here’s the kicker: once your contracted token volumes run dry, those sweet savings evaporate—faster than you can say “renegotiate or pay full price.” If you’re buying AI at scale, this isn’t just about how much you spend anymore. It’s about where and how you spend it, and being caught off guard could mean budget headaches you didn’t see coming. Intrigued? You should be. LEARN MORE

Enterprise clients chasing better rates must commit a set percentage of AI spend to Anthropic, as discounts end once contracted token volumes run out

Anthropic wants more than your business. It wants a guaranteed slice of your AI budget.

The company is requiring customers to spend a minimum percentage of their AI budget on its products before they qualify for higher discounts.

The move is part of a broader tightening of Anthropic’s enterprise pricing. It arrives as the company and its chief rival, OpenAI, compete hard for the biggest corporate accounts in the industry.

How the new terms work

The centerpiece is what the industry calls a “share of wallet” clause. Both Anthropic and OpenAI have added these clauses to their discount agreements, requiring large clients to direct the bulk of their AI spending to a single provider. In exchange, they become eligible for steeper discounts than they would otherwise get.

Advertisement

There is a practical wrinkle, though. Compliance with these clauses remains largely uncheckable. A provider can see what a customer spends with it. It generally cannot see what that customer spends with competitors.

The second piece of the policy concerns what happens when a contract’s token allocation runs out. Anthropic ends its volume discounts, which have run around 15% for large enterprise clients, as soon as those contracted token volumes are reached. At that point, customers have two options: renegotiate, or start paying standard list prices.

The OpenAI contrast

OpenAI takes a softer line on the same problem. Its policy gives clients an additional month beyond their contractual limits to renegotiate before rates revert to standard pricing.

The gap makes Anthropic’s approach the stricter of the two. Customers who blow through their token allocation faster than expected lose their discount immediately, rather than getting a buffer to sort things out.

Why Anthropic can push harder

As of June 2026, the company had more than 100 customers each spending over $10 million annually. It also counted over 1,000 customers spending over $1 million a year.

There is already at least one prominent example of churn. CodeRabbit moved its primary AI supplier from Anthropic to OpenAI. As of September 2026, CodeRabbit was spending tens of millions annually with its provider.

What this means for enterprise buyers and the AI market

For companies buying AI at scale, discounts are no longer just a function of how much you spend. They increasingly depend on how much of your total AI budget you hand to one vendor.

Share of wallet clauses ask customers to choose between spreading spending across multiple model providers or concentrating spending to earn better pricing. Because compliance is hard to verify, some buyers may treat the commitments loosely.

The token cutoff policy adds a forecasting burden. Enterprises signing with Anthropic will need to estimate their usage carefully, since overshooting their allocation triggers an immediate return to list prices.

Disclosure: This article was edited by Diego Almada Lopez. For more information on how we create and review content, see our Editorial Policy.

Post Comment