When California state Sen. Scott Wiener introduced the nation’s most ambitious AI safety bill last year, it became a flashpoint in the debate over how to govern artificial intelligence. Senate Bill 1047 caught the attention of tech billionaires, large corporations, Hollywood, researchers and the media for good reason — it aimed to establish some of the first guardrails for AI in America.
While much of the coverage centered around safety components of the bill, and was ultimately vetoed by Gov. Gavin Newsom, it advanced an important proposal that deserves continued attention. Hidden in the bill was “CalCompute,” a policy which would have established parts of the computing power driving AI development as a public resource, disrupting the lock tech giants have on this piece of technology.
Computing power — called “compute” in the industry — is the highway system of the AI revolution, the essential infrastructure required to develop and run advanced AI systems. At its core, this infrastructure consists of massive data centers filled with specialized chips that provide computers with the raw processing power needed to develop and run AI systems.
Today, this infrastructure is dominated by just three companies: Amazon, Google and Microsoft. It’s not surprising that the largest tech companies in the world have extended their dominance to the AI revolution with a close hold on compute. After all, you can only build advanced AI models if you have access to the massive computing power needed to train them, creating a barrier-to-entry for small and midsized startups and entrepreneurs.
Lawmakers can’t let CalCompute die on the vine with the veto of SB 1047.
This isn’t a new model. Public options like this are an old idea and a good one because they provide people with the means to participate in new markets and balance out structural power imbalances created by monopolies. America has invested in shared infrastructure like the U.S. Postal Service and broadband and, in turn, we have all benefited from the increased access, innovation and economic growth these public investments unleashed.
The same principle should apply to the computing power that will drive our economic future. Imagine a publicly funded, publicly run supercomputing network that serves not just government agencies and researchers, but also entrepreneurs and innovators currently priced out of the market.
This wouldn’t replace private providers — it would be complementary, expand access and ensure AI development serves both market innovation and public needs. Since this public option would be free from profit motives, it wouldn’t face pressure to prioritize commercial interests over society’s needs or users’ interests.
The stakes extend far beyond Silicon Valley. AI will reshape nearly every sector of our economy in the coming decades, from health care and scientific research to financial services and manufacturing. Three jurisdictions will play outsized roles in determining whether this transformation serves the broader public interest: Washington D.C., Brussels and Sacramento.
As home to both Silicon Valley and the world’s fifth-largest economy, California’s CalCompute proposal offers a vital model for how public computing infrastructure could work — and a template for federal action.
We’ve seen promising momentum behind this idea on the national and global stage. New York recently established Empire AI, a $400 million initiative to create public computing infrastructure through a consortium of universities. In the United Kingdom, the government’s investment in a new UK AI Research Resource aims to provide access to world-class compute for UK-based researchers. United Arab Emirates has invested heavily in cloud infrastructure.
The Omidyar Network is investing millions into the research and development of open digital ecosystems. And academics like Ganesh Sitaraman, a legal scholar at Vanderbilt University, are providing the expertise on public options for AI to ensure that the future of AI is not monopolized by already entrenched corporate players.
The window for shaping the next digital economy won’t stay open forever. Already, the biggest tech companies are spending billions to entrench their advantage in compute. But by creating public alternatives — whether at the federal level or through state initiatives like CalCompute — we can ensure that the next wave of AI innovation benefits all of us, and doesn’t fade with a governor’s veto in California.