Key Takeaways

- Y Combinator startups are receiving $500,000 in credits from both OpenAI and Anthropic, with some founders getting offers exceeding $3 million
- The credit war is happening while both companies need to improve margins ahead of expected IPOs
- Google Cloud, Microsoft Azure, and AWS are also competing with credits, early model access, and engineering support
OpenAI and Anthropic are handing out compute and token credits worth hundreds of thousands of dollars to early-stage startups. Some founders report receiving competing offers that total more than $3 million. The goal: lock them into an ecosystem before they can leave.
According to a Wall Street Journal report, Hans Ibarra, founder of AI voice startup Dialogus, has received credit offers that add up to roughly the size of an average US seed round. That's around $3 million in free infrastructure, per PitchBook data. The discount war spans AI labs and cloud providers alike, with Google, Microsoft, and Amazon all running parallel programs.
What are OpenAI and Anthropic offering?
Y Combinator companies have become the prime target. In May, Sam Altman announced $2 million in token credits in exchange for equity stakes in YC startups. Anthropic countered with $500,000 and no equity requirement. That's a significant jump from its previous $30,000 offer. OpenAI then matched the $500,000, no equity attached, plus an optional $1.5 million more in exchange for shares.
The math gets big fast. With four YC cohorts per year averaging 200 companies each, OpenAI and Anthropic could distribute up to $800 million in combined credits annually. That's real money, even for companies valued in the tens of billions.
Cursor, the AI coding tool acquired by SpaceX, offered a 75 percent discount through July 5. Google Cloud is handing out up to $500,000 in credits, early access to Gemini models, and occasional access to DeepMind engineers. Microsoft and AWS are running similar programs.
Why spend millions to acquire startups that might never pay?
Platform lock-in is the play. Once a startup builds on a particular model API, rewriting for a competitor costs engineering time and introduces risk. Credits create switching costs before the startup has revenue to spend elsewhere.
The timing is awkward. Both OpenAI and Anthropic are expected to pursue IPOs in the coming years. Public market investors will scrutinize margins. Giving away compute is the opposite of margin improvement.
But the competitive pressure is real. Open-source models and cheaper alternatives, many from Chinese labs, are eroding the pricing power of frontier models. If startups can build on Llama or Qwen at a fraction of the cost, OpenAI and Anthropic need a reason for founders to choose them anyway. Free credits are that reason.
How Anthropic caught up and OpenAI responded
Anthropic's revenue surged late last year, driven by Claude Code and its Cowork collaboration features. OpenAI reportedly didn't catch up until it released GPT-5.4 in March. Now OpenAI is actively selling its Codex tool to startups, and the credit offers are part of that push.
The escalation pattern is clear. Anthropic raised its YC offer from $30,000 to $500,000. OpenAI matched and added an equity-linked tier. Neither company can afford to lose the next batch of AI-native startups to the other.
What cloud providers are doing
Google Cloud's program includes not just credits but direct access to research staff. A spokesperson confirmed the company offers up to $500,000 in credits, early Gemini access, and occasional DeepMind engineer time. Microsoft and AWS haven't disclosed specifics but are running comparable initiatives.
For startups evaluating cloud providers alongside AI models, the bundling matters. A founder can get credits for both compute infrastructure and model inference from the same provider, simplifying the early architecture decisions.
Logicity's Take
This is a land grab dressed as generosity. OpenAI and Anthropic are buying future revenue with current credits, betting that today's seed-stage startup becomes tomorrow's enterprise customer. The interesting question is whether switching costs will hold. As model performance converges and open-source alternatives improve, the lock-in may be weaker than these companies hope. Founders should take the credits, but build abstraction layers. The free compute won't last, and neither will the current pricing dynamics.
What founders should consider
Free credits are real value, but they come with implied obligations. Equity stakes, when offered, dilute founders. Platform dependencies, once built, are expensive to unwind. The startups taking $500,000 from both OpenAI and Anthropic are hedging correctly. Those taking $3 million from a single provider should think about what happens when the credits run out.
The IPO pressure on both companies also raises a timing question. Margins will need to improve before public offerings. That means either prices rise or credit programs shrink. Founders planning 18-month runways on free tokens should model what happens if those tokens cost money in month 13.
Context on when AI spending translates to actual returns
Frequently Asked Questions
How much are OpenAI and Anthropic giving to Y Combinator startups?
Both companies are offering $500,000 in credits with no equity requirement. OpenAI also has an optional tier providing $1.5 million in exchange for equity stakes.
Why are AI companies giving away free compute?
Platform lock-in. Once startups build on a specific model API, switching costs make them likely to stay and eventually pay. It's customer acquisition through infrastructure dependency.
How do cloud provider credits compare to AI model credits?
Google Cloud, Microsoft Azure, and AWS are offering up to $500,000 in compute credits plus additional perks like early model access and engineering support. Some startups combine both cloud and model credits from different providers.
Will these credit programs continue after IPOs?
Unlikely at current levels. Both OpenAI and Anthropic need to improve margins before going public. Credit programs will probably shrink or require revenue commitments.
Need Help Implementing This?
If you're navigating AI infrastructure decisions for your startup, Logicity can connect you with advisors who have built on both OpenAI and Anthropic platforms. Reach out at hello@logicity.in.
Source: The Decoder / Maximilian Schreiner
Manaal Khan
Tech & Innovation Writer
Produced with AI assistance and reviewed by the Logicity editorial team. Learn more in our Editorial Policy.
Related Articles
More in AI & Machine Learning
Bezos AI Lab Gets $10B: What Project Prometheus Means
Jeff Bezos is closing a $10 billion funding round for Project Prometheus, an AI lab focused on physics-based AI for manufacturing and engineering. With a $38 billion valuation and backing from JPMorgan and BlackRock, this signals a major shift in enterprise AI investment toward industrial applications.

Kimi K2.6 Open-Weight AI: 300 Agents at a Fraction of the Cost
Moonshot AI's Kimi K2.6 matches GPT-5.4 and Claude Opus 4.6 on coding benchmarks while running 300 parallel agents. For businesses locked into expensive API contracts, this open-weight model could slash AI infrastructure costs while delivering enterprise-grade automation.




