OpenAI has just spent $70 billion to buy back a large number of shares from its employees.
According to Bloomberg, the buyer for this employee stock repurchase is OpenAI itself rather than external investors. The repurchase targets include current and former employees, with a transaction valuation of $8520 billion, consistent with the valuation anchored in the $1220 billion funding round in March this year.
https://www.bloomberg.com/news/articles/2026-08-10/openai-buys-back-7-billion-of-employee-shares-in-tender-offer
The company secretly filed for an IPO with the US Securities and Exchange Commission in June but chose to provide a cash-out channel for employees with its own funds first instead of handing it over to the secondary market. The signal is clear: the listing may not happen so quickly.
Coincidentally, on the same day, August 11, OpenAI lost another executive — Brad Lightcap announced his departure on X, planning to start his own venture.
https://x.com/bradlightcap/status/2087211567012032862
He had been at OpenAI for 8 years, serving as COO (Chief Operating Officer), and was the architect of the company's commercial system. He is the latest name on OpenAI's list of executive departures in 2026.
From Power to Departure
Brad Lightcap's 4 Months
When Lightcap joined in 2018, OpenAI didn't even have GPT-2 yet.
In the following years, he built the commercial framework of the company single-handedly: finance, legal, human resources, enterprise security, government affairs, partnership relations—the initial versions of these teams were all established by him.
After being promoted to Chief Operating Officer in 2022, he was seen as Altman's "secret weapon": media copyright negotiations, enterprise partnerships, fundraising, infrastructure—most of the tedious but crucial tasks beyond the models fell on his shoulders.
In November 2023, when Altman was suddenly dismissed by the board, Lightcap was the one who stepped forward the next day to send an internal letter to the entire company to stabilize morale.
In March 2025, Altman handed over daily operations, global expansion, and commercial strategy to him, freeing himself up to focus on R&D and products.
He effectively became the number two person on the commercial side.
The turning point came a month later.
In April 2025, OpenAI restructured its management, and Lightcap relinquished his title of Chief Operating Officer, moving to oversee special projects, mainly handling complex deals and investments.
According to The Information, this role was already distant from daily organizational management.
https://www.theinformation.com/briefings/openai-special-projects-lead-former-coo-brad-lightcap-depart
From gaining power to being sidelined, it took only one month.
Four months later, he announced his departure.
In his farewell letter, he wrote: "For the past few months, I've been thinking about what the next frontier is and what might hinder the success of the mission. I believe there are some important and entirely new things that the world needs to address."
He did not reveal specific plans.
Altman was also quick to bid him farewell on X, saying he took on every function and challenge OpenAI needed, "We wouldn't be where we are today without you."
https://x.com/sama/status/2087243765765509185
A person who experienced 8 years at OpenAI, a power struggle, and multi-billion dollar financing chose to leave as the company's valuation reached $8520 billion and an IPO approached.
What did he see? There is no answer yet.
9 Executives Left in Half a Year
Lightcap is not an isolated case.
From April to August 2026, at least 9 OpenAI executives or key role holders have departed.
In April, four left first: former Chief Product Officer Kevin Weil, Sora lead Bill Peebles, and Chief Technology Officer for Commercial Applications Srinivas Narayanan left around the same time. Chief Marketing Officer Kate Rouch stepped down due to needing time for cancer recovery.
July saw a higher density: Application Business Lead Fidji Simo shifted to a part-time advisor role due to chronic illness. Security Systems Lead Johannes Heidecke, Chief Futurist Joshua Achiam, and Ethics Lead Chloé Bakalar left successively.
In August, Lightcap added the latest line.
ParaFi partner Jeff Park's comment on X is thought-provoking:
It's very unusual for so many executives to leave right before a highly anticipated IPO—unless...
He didn't finish, deciding to be cryptic—"Those who understand, understand; those who don't, there's no need to explain."
https://x.com/dgt10011/status/2087247008822165592
The Only Ethics Officer Left, No Successor Appointed
Among the list of 9, Bakalar's departure was the quietest and also the most worthy of separate mention.
According to the UK's Financial Times, Bakalar joined OpenAI as Ethics Lead last August. Previously, she was Meta's Chief Ethics Officer, spending six years establishing AI ethics systems for Instagram and Facebook.
She left in July this year without any public statement.
According to sources cited by the Financial Times, she was the company's only dedicated ethics researcher, and no successor was appointed after her departure—OpenAI no longer has a full-time ethics overseer.
https://www.ft.com/content/e49dfb75-f841-4466-a577-f7aaff8779a0
The OpenAI spokesperson's response was:
AI ethics does not belong to one person or one team. Ethical considerations are deeply embedded into the model-building process driven by multiple research teams.
Bakalar herself said something similar earlier this year at an AI conference: ethics is everyone's responsibility, and there shouldn't be just one person serving as the moral center for an AI developer.
She also said:
We've been building the plane while flying it.
The timing of her departure is delicate.
Just in recent weeks, OpenAI admitted its models, during testing, unauthorizedly hacked into Hugging Face's servers. The US subsequently introduced release restrictions for powerful models. OpenAI announced last week that it would delay the release of its next-generation model Astra to strengthen safety controls.
With the Security Systems Lead, Chief Futurist, and Ethics Lead all vacant in the same month, OpenAI's Chief Research Officer Mark Chen told Wired last month that the company is integrating safety work into the frontier model development process, allowing safety teams to participate earlier in model and release decisions.
Delayed IPO and Catching-up Anthropic
The $70 billion employee stock repurchase is a compromise: employees can cash out without waiting for the IPO, and the company reduces the urgency of going public.
The IPO documents OpenAI secretly filed in June are still valid, but The New York Times reported that the company prefers to delay until next year.
Altman admitted last month:
The past 12 months have not been our best year, largely my responsibility.
The Wall Street Journal reported in April that the company had missed internal financial targets.
The external competitive landscape is also changing dramatically.
Anthropic reportedly became profitable earlier this year, its valuation surpassing OpenAI. Prediction market Myriad currently generally believes Anthropic will go public before OpenAI this year.
https://myriad.markets/events/which-companies-will-ipo-before-2027
Bloomberg believes that Anthropic has strong momentum in AI software, its valuation has already surpassed OpenAI, and it is poised to go public first.
The departing executives have already begun their next steps.
According to Business Insider, OpenAI's former CPO (Chief Product Officer) Kevin Weil is building an AI science startup, seeking a valuation of at least $750 million.
https://www.businessinsider.com/weil-valuation-750-million-for-ai-science-startup-2026-8
The plan of the latest departure, Lightcap, has not been announced, but he hinted in his farewell letter that the direction is related to "factors hindering the success of the mission."
These people have seen the full picture of AI at OpenAI: models, business, capital, safety.
They chose to leave on the eve of the IPO to do what they consider more important. These choices themselves are rich with information.
OpenAI's IPO timeline will be the next key node to test the actual impact of this wave of turbulence.
References:
https://x.com/bradlightcap/status/2087211567012032862
https://techcrunch.com/2026/08/10/openai-reportedly-completed-a-7-billion-employee-tender-offer/
https://www.ft.com/content/e49dfb75-f841-4466-a577-f7aaff8779a0
This article is from WeChat public account "New Zhiyuan", author: ASI Apocalypse, editor: Mark
