The $23M Bet: Inside the National AI Academy

In July 2025, three of the most competitive companies in artificial intelligence did something that looked, on the surface, like a category error. OpenAI, Microsoft, and Anthropic — rivals who spend billions trying to out-ship one another — jointly agreed to fund the same project, run by the same teachers' union. The project is the National Academy for AI Instruction, a partnership with the American Federation of Teachers (AFT) and its New York City affiliate, the United Federation of Teachers (UFT).

The number attached to it is $23 million. The goal attached to it is not a product feature or a model benchmark. It is people: roughly 400,000 U.S. educators trained in AI by 2030 — about one in every ten teachers in the country. This article is the map to what that money is buying, who put it up, and why the timing matters.

Key Takeaways

• The Academy is a $23M training initiative, not a product. It teaches educators how to use AI; it does not sell them a chatbot. That distinction is the whole point.
• Three rivals funded it together. OpenAI contributed $10 million and Microsoft $12.5 million, with Anthropic also contributing, according to the AFT's announcement — the first time a national teachers' union has partnered with multiple AI companies at once.
• The target is 400,000 teachers by 2030. That is roughly 10% of the U.S. teaching workforce, reached through a flagship New York City campus that is meant to scale nationally.
• It predates the free ChatGPT product. The Academy was announced in July 2025, months before "ChatGPT for Teachers" launched — they are two arms of the same education strategy, not the same thing.
• The training layer is where trust is won or lost. Whoever teaches 400,000 teachers to think about AI shapes how the next decade of classrooms actually use it.

What Is the National Academy for AI Instruction?

Strip away the press-release language and the Academy is a professional-development operation. Its product is workshops, courses, and hands-on training that help K-12 educators understand what AI tools do, where they help, and where they fail. The AFT describes it as the first partnership of its kind between a national teachers' union and a group of leading AI companies, built to create sustainable infrastructure for AI in education rather than a one-off training day. You can read the union's own framing in its launch announcement.

The word that matters most in that sentence is infrastructure. A single vendor demo teaches you one tool. An academy — with a physical campus, a curriculum, and a multi-year funding commitment — is designed to outlast any single model release. When GPT-5.1 gives way to GPT-6, or when a district switches vendors, the underlying skill of evaluating and directing AI is supposed to remain.

That framing is why the Academy is worth understanding on its own terms, separate from any free-account offer. It is an attempt to build a durable answer to a question every school in America is now asking: how do we get our staff fluent in this technology without betting the whole building on a single company?

Who Funded It, and How Much?

Here is the part that makes people stop scrolling. Coverage from EdWeek and TIME laid out the contributions:

Two facts jump out of that table. First, every funder also sells a competing classroom product. Microsoft has Copilot for Education, OpenAI has the free ChatGPT for Teachers offer that this very series covers, and Anthropic runs Claude for Education — which this site tracks in its sibling Claude for Teachers series. Second, they still pooled money into a shared training effort run by a neutral third party. Why three competitors would do that is a genuinely interesting question, and it is the subject of the next piece in this series, Why OpenAI, Microsoft, and Anthropic Fund One Union.

Why $23 Million, and Why Now?

The size of the check tells you the funders are treating teacher training as a strategic bottleneck, not a charity line item. Adoption of any classroom AI tool stalls at the same place: a teacher who does not trust the output, or does not know how to steer it, quietly stops using it. Twenty-three million dollars aimed at removing that friction is a bet that the constraint on AI in schools is human confidence, not model capability.

The timing is not an accident either. The Academy was announced in the summer of 2025, during a window when districts were swinging between outright bans and cautious pilots, and when public skepticism about AI in classrooms was loud. Getting a trusted, teacher-led institution to own the training narrative — before a fragmented, vendor-by-vendor training market could form — is worth a lot to companies whose long-term markets depend on schools saying yes. Outlets like K-12 Dive covered the launch precisely because the union, not a tech company, was placed at the center of it.

How does this connect to the free ChatGPT offer?

It is easy to conflate the two, so be precise: the Academy trains teachers on AI in general, while "ChatGPT for Teachers" is OpenAI's specific free product for verified U.S. K-12 educators through June 2027 (unlimited GPT-5.1 Auto, search, file uploads, connectors, and image generation, with data not used to train models by default). The Academy came first and is broader. The product is one company's play inside the environment the Academy helps create. If you want the ground-level product details, start with what's actually free and for whom.

What Should a Teacher Actually Do With This?

For an individual educator, the Academy is less something you sign up for tomorrow and more a signal about where the ground is moving. The practical move is to build the skill it is trying to teach — directing AI well — starting with the tasks that eat your week. Lesson planning is the obvious first rep; a structured helper like Lesson Plan Studio turns a blank page into a draft you edit rather than author from scratch. For a broader set of classroom-ready tools, the education agent skills collection is a reasonable place to browse.

The bigger takeaway is one of leverage. The pace of that rollout — 400,000 teachers in five years — is ambitious enough to deserve its own scrutiny, which is exactly where this series goes next.

Frequently Asked Questions

Is the National Academy for AI Instruction the same as ChatGPT for Teachers?

No. The Academy is a $23M, multi-funder training initiative run with the AFT and UFT. ChatGPT for Teachers is OpenAI's separate free product. The Academy was announced first and covers AI broadly, not one company's tool.

How much did each company give?

Reporting attributes $12.5 million to Microsoft and $10 million to OpenAI, with Anthropic also contributing an amount that was not disclosed in the coverage reviewed here. The total is described as roughly $23 million.

How many teachers is the Academy trying to reach?

About 400,000 U.S. educators by 2030 — close to one in ten teachers nationally — starting from a flagship campus in New York City that is expected to scale across the country.

Why would competitors fund the same project?

Because teacher trust and AI fluency are a shared bottleneck. All three companies benefit if educators feel confident using AI, even as they compete on the individual products. The strategic logic gets its own article in this series.

Is it free for teachers to participate?

The Academy is funded by the companies and run through the union, and reporting frames it as training built for educators rather than a paid course. Program specifics are set by the AFT and UFT; check the union's channels for enrollment details as the NYC campus scales.

Part 15 of 100 in the ChatGPT for Teachers series. Previously: What a K-12 AI Data Breach Would Actually Mean. Next: Why OpenAI, Microsoft, and Anthropic Fund One Union. Browse more builder insights or explore AI skills for education at aiskill.market.