Anthropic announced an expanded Claude for Startups programme on 6 October, as part of SF Tech Week. Eligible companies, founded in the last five years or funded in the last two, could claim a free year of Claude Team for up to five users plus $1,000 in API credits, alongside a wider partner package worth up to $45,000.
Three days later the free Team and credit offers were paused. Anthropic said applications in the first 48 hours were 21 times those of the previous five months, and its head of startup marketing admitted on X that the acceptance criteria were not tuned properly and “way too many people” got in. Applicants who have not claimed benefits are being asked to re-verify, and some previously approved applicants may be denied. The partner discounts remain. The reversal drew open criticism from founders.
Why this matters beyond the drama
The numbers are a signal in themselves. Hundreds of thousands of people applied for free AI credits within days. That is how much of the market is now trying to build something with AI, and why cheap or free access is being rationed.
For a founder, the lesson is about planning. Free credits are marketing, and marketing budgets get cut with no notice. If your product plan assumes a vendor credit will cover your first year of AI costs, you have a hole in it.
What to do instead
- Budget AI usage at list price from day one, and treat any credit as a bonus.
- Keep the model swappable. A well-built product can move between providers without a rewrite. That is cheap to design in early and expensive to retrofit.
- Know your unit cost. Work out what one customer action costs in model usage before you set a price.
So what
If you are scoping an AI product, ask your development partner how the architecture handles a change of model, price or provider, because this month has shown all three can change within days. Our AI product development work builds that flexibility in from the start. If you want your plan stress-tested, get in touch.