OpenAI's monetization transformation is blocked, and executives are worried that California's review pressure may cause $19 billion funding to run aground

The Zhitong Finance App learned that some media reported, citing information revealed by people familiar with the matter, that OpenAI executives are concerned that rising political scrutiny pressure in California may hinder its efforts to successfully transform into a for-profit company, and have discussed moving OpenAI headquarters out of the state as one of the last resort for the transformation of the profit program.

According to the report, some of California's largest charitable organizations, non-profit organizations, and labor groups are joining hands to oppose the large-scale restructuring plan of this AI superunicorn. Since US tech giant Microsoft (MSFT.US) invests heavily in OpenAI is actually controlled by a non-profit organization, they are asking the state's attorney general to ensure that a new company they created does not violate the state's charitable trust law.

OpenAI did not immediately respond to media requests for comment. However, an OpenAI spokesperson told the media that the company currently has no plans to leave California.

According to media reports, the California and Delaware attorneys general are reviewing the proposed plan proposed by OpenAI. Regulators have a legal responsibility to protect charitable causes in the state, and they have the power to sue OpenAI for possible violation of nonprofit laws or require the AI startup to pay huge settlements under the terms of the restructuring.

OpenAI, led by CEO Sam Altman, currently operates under a structure that does not issue traditional shares and is controlled by a non-profit parent company. According to media reports, this structure is unpopular among its investors and potentially large co-criminal organizations interested in investing in OpenAI, so Altman and Microsoft, the majority shareholder of OpenAI, are pushing for changes.

Citing sources familiar with the matter, the report said that OpenAI's funders are close to $19 billion — about half of the startup's total funding over the past year — but with the added condition that shares in the new for-profit company must be obtained. If the restructuring cannot be carried out, they may withdraw this huge amount of funding, which will potentially hinder OpenAI's ambition to build large-scale data centers, build customized high-performance AI ASIC chips, and stay at the cutting edge of AI research for a long time. This can be described as directly related to OpenAI's large-scale model training, data centers, and the $10 billion AI ASIC chip plan. Because of this, OpenAI's senior management is passively in the midst of “financing claims versus public welfare compliance.”

The “nearly $19 billion” here is not revenue or valuation, but rather the amount of financing promised or arranged by the funders in the past year (about half of the total financing for that year), but the prerequisite is that OpenAI completes the restructuring, establishes a new for-profit company, and allows these investors to obtain shares in the new company. If the restructuring fails, the relevant funds may not be available or withdrawn.

The report quoted people familiar with the matter as saying that when OpenAI executives first announced the restructuring plan last year, they did not expect such a strong reaction. The findings of the California Attorney General's investigation have been a major concern in recent months. The report pointed out that if the California Attorney General complicates the restructuring, OpenAI has discussed the possibility of moving out of California at the last minute. However, the move will be very difficult because its AI research team is highly concentrated in San Francisco.

“We will continue to work constructively with the California and Delaware Attorney General's offices,” an OpenAI spokesperson said. He added that the company wants to build one of the most resourced non-profit organizations in history.

Why is it so difficult for OpenAI to completely transform into a for-profit organization?

OpenAI's “nonprofit + capped profit (capped-profit)” hybrid structure, which is fully controlled by the non-profit parent body, combined with the strict supervision of “charitable assets must not be misused by private interests” under the California/Delaware Charity Act, the intervention of the two state attorneys general in the review, and the investor's binding of nearly $19 billion to “taking shares in a new company” forms a triple legal, governance, and financing constraint: once control/core assets are transferred from the non-profit side to the for-profit entity, it is easy to trigger the red line of charitable trust/public interest compliance; On my own In 2025, they also publicly promised “continued non-profit control”, further tightening the viable path.

Since 2019, OpenAI has adopted the “OpenAI Nonprofit (501 (c) (3) charity parent) + OpenAI LP/Global (profit arm capping profit)” architecture, and the parent body controls the profit arm through OpenAI GP LLC; the design aims to “both raise capital and not deviate from the charitable mission”. This set of “capping profit” and the logic of parent control is written in the official structure description and 2019 announcement.

From system design to public promises, OpenAI has always bound “non-profit control + public welfare mission” to the company's “steering wheel”; changing to a standard corporate system with for-profit shareholders as the ultimate beneficiaries is tantamount to loosening this “steering wheel” — involving unbinding the triple restrictions of law, governance, and financing.

In California, the Attorney General has strong control over the preservation and use of charitable assets. Transactions involving disposal or transfer of control of all or substantially all assets require prior notification/review to prevent “privatization of public welfare assets”. Once the core IP/business moves from the non-profit side to the for-profit side, it is very easy to trigger a “Charitable Trust Act/Public Interest” review. Therefore, at the legal level, which is critical to the restructuring of OpenAI, non-profit assets and missions are “locked in”; in terms of governance/public opinion, the “public interest first” hard constraint is constantly being strengthened. In addition to fair pricing, handing over control or significant assets to for-profit companies usually requires regulatory permit/court approval, and faces a high risk of litigation.

Hire a close Newson friend to help with the transformation plan

According to media reports, the startup recently hired advisors close to California Governor Gavin Newson (including former Senator LaFonza Butler) to solicit support for the program in Sacramento. The company held statewide listening sessions with civil society groups throughout the summer and pledged $50 million to support nonprofit and community organizations.

In May of this year, OpenAI stated that the non-profit entity that controls OpenAI will continue to have control after it turns into a for-profit company. OpenAI Chairman Brett Taylor said in a statement at the time: “After listening to civic leaders and having constructive conversations with the Delaware and California Attorney General's offices, we have decided that a nonprofit organization will continue to control OpenAI.”

However, political scrutiny is intensifying. The report said that on Friday, California and Delaware attorneys general sent a letter to OpenAI to express concern about its safety commitments in the context of recent reports of long-time chatGPT interactor suicides (including a murder-suicide in Connecticut last month).

According to reports, part of the negotiations between OpenAI and the California Attorney General revolved around ensuring that the non-profit organization has independent control over the new company. Regulators see these suicides as a sign that OpenAI is putting the use and revenue of ChatGPT above its mission to “create AI for the public good,” the report cites a person familiar with its work.

OpenAI said it is addressing issues related to “flattering” and flattering (that is, the programming of the AI system causes it to cater too much to users), and will launch a parental control function for ChatGPT. According to reports, OpenAI Chairman Brett Taylor said, “We will do our best to respond to the concerns of both state attorneys general.”

OpenAI was founded in 2015 as a non-profit organization, then created for-profit subsidiaries, which enabled it to raise tens of billions of dollars from Microsoft and venture capital agencies.

In July of this year, it was reported that Microsoft is in in-depth negotiations on an agreement that will allow it to continue to obtain key OpenAI technology, which will remove a major obstacle to the transformation of OpenAI into a for-profit organization.

OpenAI is also dealing with a lawsuit brought by Musk and its rival AI startup xAI to block this transformation and accuse OpenAI of violating its non-profit mission. The case is expected to be heard next year. OpenAI said the lawsuit was “baseless and unsupported by facts.”

Altman and OpenAI's AI ambitions

OpenAI's helmsman, CEO Altman, recently said in an interview with the media that OpenAI will invest trillions of dollars in the “near future” for AI's core infrastructure, including AI chips, high-performance data center network equipment, advanced power systems, and other underlying resources that support the operation of increasingly huge AI training/inference systems. Altman believes that in the long run, AI will be the most important thing in the world, and it is impossible for the whole society as a whole to regret investing heavily in AI.

As of early August, ChatGPT, developed by OpenAI, now has over 700 million weekly active users globally — far more than 500 million in March. Additionally, current and former employees of OpenAI plan to sell shares worth around $6 billion to an investor group made up of Thrive Capital, SoftBank, and Dragoneer investment giants. The deal will bring the ChatGPT developer's valuation to $500 billion — a valuation that also means the world's strongest unicorn is born.

It can be said that $500 billion far exceeds Elon Musk's SpaceX valuation. OpenAI management expects revenue to triple to $12.7 billion this year, compared to around $3.7 billion in 2024.

Last week, some media reported, citing information revealed by people familiar with the matter, that ChatGPT developer OpenAI will cooperate with American chip giant Broadcom (AVGO.US) and plans to launch OpenAI's first customized artificial intelligence chip (the so-called AI ASIC chip) next year. It is estimated that the order size of this cooperation between the two parties will exceed 10 billion US dollars.

For OpenAI, which developed ChatGPT, an epoch-making AI application, it relies on astronomical computing power resources when training and running its artificial intelligence system. This is also an important logic that the AI unicorn, which has a valuation of up to 500 billion US dollars, has continued to spend huge sums of money to buy the Nvidia H100/H200 and the newly launched Blackwell architecture AI GPU in recent years. The recent announcement to jointly build an AI ASIC computing power cluster with Broadcom also means that, according to the AI unicorn, it is much more cost-effective to jointly build AI computing power infrastructure with the technical path of customized AI chips compared to simply stacking up Nvidia AI GPUs.

As demand for the AI computing power infrastructure required to train and run large AI models continues to surge, OpenAI's initiatives have followed the AI ASIC steps of Google (GOOGL.US), Amazon (AMZN.US), and Facebook parent company Meta (META.US). These tech giants have all built customized AI chips to handle extremely large AI workloads.