OpenAI's 90% discounts are buying customers, not earning them

OpenAI is handing out discounts of up to 90% to pull enterprise customers away from Anthropic. That's not a pricing strategy. That's a land grab — and if you're a business leader signing one of these deals this quarter, you should know exactly what you're buying into.

The Decoder, citing reporting by The Information, published details this week of the campaign. The numbers are striking. DataCamp CEO Jonathan Cornelissen said OpenAI's pricing cut his company's costs by 80 to 90 percent. Replit switched to OpenAI after being offered more than 30% off a contract worth over $100 million a year. A US bank with more than $7 billion in assets had its ChatGPT seats doubled to 500 employees at no extra charge — paying under $7 per user instead of the usual $20 or more.

Rippling CEO Parker Conrad put the shift in blunt terms: his company moved OpenAI's share of its coding workload from 16 to 52 percent, while Anthropic's share fell from 57 to 18 percent. Anthropic's overall growth reportedly slowed from Q2 to Q3, according to an investor memo cited by The Decoder. The campaign appears to be working.

This isn't just efficiency — it's margin sacrifice

To be fair, OpenAI hasn't conjured these discounts purely out of desperation. The Decoder notes that the company has achieved real cost reductions through its Sol and Luna model lines, driven by more efficient models and infrastructure. So some of the price drop reflects genuine engineering progress, not just a checkbook fight.

But here's the catch: it's not clear how much of the discount is efficiency and how much is margin sacrifice. Those two things have very different implications for anyone signing a multi-year contract today. If the cuts are efficiency-driven, prices could stay low. If they're subsidized to win share, they won't hold once the competitive pressure eases — or once OpenAI's investors want returns.

The strongest counterargument is simple: cheap is cheap. If DataCamp's costs genuinely fell 80 to 90 percent and the models perform, why should it care about OpenAI's unit economics? That's a reasonable position for a buyer. The problem is that predatory pricing — selling below sustainable cost to eliminate a competitor — tends to end one of two ways: the aggressor runs out of runway, or the competitor folds and prices snap back. Neither outcome is great if you've rebuilt your stack around the winner's API.

And both OpenAI and Anthropic are facing a third pressure that makes this whole fight look even more precarious. The Decoder points out that Chinese models from Deepseek and Kimi are delivering comparable performance for a fraction of the cost. OpenAI isn't just trying to beat Anthropic — it's trying to justify premium Western pricing against a wave of cheap, capable alternatives. Slashing prices to win enterprise accounts makes sense as a short-term tactic. As a long-term moat, it's thin.

What you're actually betting on

If you sign a discounted OpenAI deal today, you're making a specific wager: that OpenAI has enough capital and operational discipline to sustain these prices long enough for switching costs to set in, and that Anthropic either can't match the pricing or fades as a serious option. That might be right. OpenAI has significant backing and a larger consumer brand. But it's a bet, not a sure thing, and you should treat it like one.

The businesses most exposed are the ones that don't just buy API access but rebuild workflows, fine-tune models, or train internal teams around a specific provider's tooling. Once your engineering org has reorganized around one vendor's SDK and your evals are calibrated to its outputs, switching isn't a procurement decision anymore — it's a project.

What it means for you

If you're being approached with a heavily discounted OpenAI offer right now, three things are worth doing before you sign. First, ask your legal team what the contract says about price changes at renewal. A 90% discount that reverts to list price after 12 months isn't a deal — it's a trial with a cliff. Second, check what your current Anthropic contract says about early termination; the savings from switching may not survive the exit costs. Third, think hard about how deeply integrated you'll be after migration. If the answer is 'very,' the discount needs to be large enough to justify the lock-in risk, not just the current invoice.

My bet: OpenAI wins enough of these accounts to slow Anthropic's growth further in Q4, but the pricing won't hold at these levels past mid-2027. Watch what happens at renewal time for the first cohort of customers who signed deals like Replit's. That's when we'll find out whether this was a sustainable strategy or an expensive audition.