Open AI’s Sam Altman: Building a strategic national reserve of computing power makes a lot of sense
**Clarifying OpenAI’s Position on Government Involvement and AI Infrastructure**
There have been some recent questions and concerns about OpenAI’s approach to government involvement in our infrastructure, and we’d like to clarify our position.
**No Government Guarantees for OpenAI Data Centers**
First and foremost, we do not have — nor do we want — government guarantees for OpenAI data centers. We firmly believe that governments should not pick winners or losers in the marketplace, and taxpayers should not be asked to bail out companies that make poor business decisions or lose in the market. If one company fails, others will step up and continue doing good work.
**Government-Owned AI Infrastructure**
What we do see as potentially sensible is governments building (and owning) their own AI infrastructure. In such cases, the benefit and upside of those assets should flow to the government, not private companies. We can envision a world where governments decide to secure large amounts of computing power and determine how that capacity is utilized. In these situations, it might make sense for governments to obtain a lower cost of capital to fund such initiatives. Building a strategic national reserve of computing power is a logical step, but again, these resources should be for the government’s own purposes.
**Loan Guarantees and Semiconductor Fabs**
The only area where we have discussed government loan guarantees is in support of the buildout of semiconductor fabs in the US. Along with other companies, we have responded to the government’s call regarding building an independent chip supply chain to bring jobs, industrialization, and strategic strength back to the US. However, we did not formally apply for these guarantees. Importantly, this is very different from the government guaranteeing the construction of data centers for private benefit.
**Addressing Key Questions and Concerns**
We understand that there are a few underlying questions driving this discussion:
**1. How Will OpenAI Pay for Its Infrastructure?**
We expect to end this year with an annualized revenue run rate above $20 billion and aim to grow to hundreds of billions by 2030. We are currently considering commitments of about $1.4 trillion over the next eight years. This level of investment obviously requires significant and sustained revenue growth, and achieving each milestone will require considerable effort.
We are optimistic about our prospects, especially with upcoming enterprise offerings, new consumer devices, and categories like robotics. We also foresee new opportunities, such as AI-driven scientific discovery. Additionally, we are exploring ways to sell compute capacity directly to other companies and individuals — the need for “AI cloud” services is only going to grow. We may raise more equity or debt capital in the future to support this growth.
Everything we see suggests that the world will need much more computing power than what we’re currently planning, and we want to be ready to meet that demand.
**2. Is OpenAI Trying to Become “Too Big to Fail”?**
Absolutely not. If we make mistakes and can’t recover, we should fail — and other companies will continue serving customers and driving innovation. That’s how capitalism should work, and the broader ecosystem and economy will be fine. Our goal is to build a successful company, but if we get it wrong, that’s on us.
There has been some discussion about government financing. To clarify, we believe the US government should have a national strategy for its own AI infrastructure, separate from private firms. As a recent example, Tyler Cowen asked about the federal government acting as “insurer of last resort” for AI risks, akin to nuclear power. My comment was that, yes, government would ultimately bear responsibility in the case of catastrophic misuse (such as a rogue AI-enabled cyberattack), but this is entirely different from bailing out a company or insuring against business failure. We do not believe the government should be insuring AI companies.
**3. Why Invest So Much Now Instead of Growing Slowly?**
We are trying to build the infrastructure required for a future economy powered by AI. Given our research and the trends we observe, now is the time to invest and scale up our technology. Massive infrastructure projects take years to build, so we have to start early.
Currently, severe compute constraints lead us — and others — to rate limit our products and delay new features and models. In a future where AI can make significant scientific breakthroughs but requires tremendous computing power, we want to be prepared for that moment. The risk of not having enough compute capacity is currently greater than the risk of having too much.
Our mission is to accelerate the beneficial impacts of AI on humanity, including advancing cures for deadly diseases and eventually bringing Artificial General Intelligence (AGI) to more people as quickly as possible. We anticipate massive demand for affordable, abundant AI, and believe this technology could improve lives in countless ways. While we are taking bold steps, if we are wrong, the market — not the government — will handle the consequences.
**Conclusion**
We are privileged to be in a position to help build the future of AI and are making our bets based on what we see from our vantage point. We may not always get it right, but we are committed to transparency and responsible growth — with private risk, not government guarantees.
https://www.shacknews.com/article/146718/openai-ceo-sam-altman-government-ai
You may also like
Leave a Reply