History & Progress

CalCompute's origins trace to SB-1047, the Safe and Secure Innovation for Frontier Artificial Intelligence Models Act, introduced by State Senator Scott Wiener on February 7, 2024. SB-1047 was the first major attempt by any U.S. state legislature to comprehensively regulate frontier AI development — applying to models trained using more than 10²⁶ floating-point operations at a cost exceeding $100 million.

SB-1047's key provisions would have required frontier AI developers to write and publish safety protocols, conduct pre-deployment risk assessments, maintain full shutdown capability, submit to annual third-party audits, and protect whistleblowers who report safety violations.

SB-1047 also included the original legislative proposal to create CalCompute — a public cloud computing cluster associated with the University of California, intended to give startups, researchers, and community groups access to large-scale compute resources they could not otherwise afford.

The bill passed the State Senate 32–1 on May 21, 2024. Following significant amendments in August 2024, it passed the Assembly 48–16 and the Senate again 30–9 on August 28–29, 2024.

Support & Opposition

SB-1047 attracted significant national attention. Notable supporters included AI pioneers Geoffrey Hinton and Yoshua Bengio (both Turing Award recipients), Stuart Russell, Lawrence Lessig, Scott Aaronson, Max Tegmark, and OpenAI whistleblowers. Organizational supporters included the Economic Security Project and ENCODE — both now partners of this coalition. Over 120 Hollywood figures also signed in support.

Notable opponents included AI researchers Andrew Ng, Fei-Fei Li, and Yann LeCun, as well as members of Congress including Nancy Pelosi and Ro Khanna, who argued the bill was overly broad and could stifle open-source AI development.

The Veto

On September 29, 2024, Governor Gavin Newsom vetoed SB-1047. His stated rationale was that the bill's regulatory framework targeted AI models based solely on their computational size without accounting for whether models were deployed in high-risk environments. Simultaneously, Newsom announced plans to convene a working group of leading AI researchers — including Dr. Fei-Fei Li, Justice Mariano-Florentino Cuéllar, and Dean Jennifer Tour Chayes — to develop an empirical, science-based framework for AI governance. The window for the Legislature to override the veto closed on November 30, 2024, without action.

Following the veto of SB-1047, Senator Wiener returned in the 2025 legislative session with SB-53, a narrowed but still consequential AI bill. Where SB-1047 sought to regulate the full lifecycle of frontier AI development, SB-53 focused on transparency, whistleblower protections, and catastrophic risk reporting for frontier AI developers — and critically, preserved and formalized CalCompute as a standalone initiative.

Earlier in 2025, the working group convened by Governor Newsom following his SB-1047 veto — led by Dr. Fei-Fei Li, Justice Mariano-Florentino Cuéllar, and Dean Jennifer Tour Chayes — published a landmark working report providing an empirical, science-based framework for California AI governance. SB-53 was designed to be responsive to that report's recommendations.

SB-53 was signed by Governor Newsom on September 29, 2025 — exactly one year after he vetoed SB-1047 — and enacted as Chapter 138 of the California statutes. Under SB-53, CalCompute is established in California Government Code Section 11546.8 as a public cloud computing cluster to be developed by a 14-member Consortium housed within GovOps. The Consortium must deliver a framework report to the Legislature by January 1, 2027.

"With a technology as transformative as AI, we have a responsibility to support that innovation while putting in place commonsense guardrails to understand and reduce risk. With this law, California is stepping up, once again, as a global leader on both technology innovation and safety."

— Sen. Scott Wiener, on the signing of SB-53