OpenAI announces that 14 core executives have departed in 2026.

As of August 26, 2026, publicly available information confirms that at least 14 senior or key leaders at OpenAI have left their roles this year, spanning revenue, product, science, safety, ethics, marketing, sales, hardware, and data centers.

Author and source: 0x9999in1, ME News



• As of August 26, 2026, publicly available information confirms that at least 14 senior or key leaders at OpenAI have left their roles this year, spanning revenue, product, science, safety, ethics, marketing, sales, hardware, and data centers.
• This is not a “mass exodus.” Multiple factors—including entrepreneurship, health reasons, business contraction, team restructuring, and governance disagreements—are at play; attributing all departures solely to a loss of confidence in OpenAI is not rigorous.
• What the market should truly be wary of is not the number 14 itself, but the fact that these departures are concentrated in key areas critical to a company preparing for public markets: commercialization, product, risk control, and computing infrastructure.
• OpenAI’s business has not collapsed. The company reports that its weekly active users have exceeded 1 billion and its enterprise customers have surpassed 2 million; public reports in August showed its annualized revenue run rate has exceeded $40 billion. The issue is precisely this: the larger the business, the higher the cost of organizational disorder.
• OpenAI secretly filed its IPO documents with the U.S. SEC in June, but the latest public information suggests the listing window is more likely to be pushed to 2027. This means it still has time to refresh its leadership, but it must also demonstrate before any roadshow that its management team can stabilize.
• Compared to Anthropic, OpenAI’s current weakness is not in storytelling, but in “organizational predictability.” As of the end of July, Anthropic’s annualized revenue run rate has exceeded $65 billion, and there has been no equivalent density of key leadership changes publicly disclosed—this affects how investors assign governance premiums or discounts to the two companies.

First, let’s clarify the most emotionally charged point: OpenAI’s recent leadership changes this year have been significant, but they cannot be simplistically described as “14 executives collectively quitting.”

As of August 21, Business Insider compiled a list of 13 departing senior executives and key personnel; on August 26, The Wall Street Journal further revealed that data center head Chris Malone had also left. Combining these two public reports results in a minimum list of 14 individuals.

Among them, Denise Dresser announced her departure as Chief Revenue Officer in August, and OpenAI has appointed Dali Rajic as her successor; Brad Lightcap left the company after eight years to “start something new,” having previously transitioned from COO to lead special projects; Fidji Simo stepped down from her full-time role to become a part-time advisor due to worsening chronic illness; Kevin Weil left OpenAI for Science to start a startup; Bill Peebles, head of Sora, departed in April; and Srinivas Narayanan, CTO of Enterprise Applications, also left in April.

Looking further down, the reasons become even more varied. Chief Marketing Officer Kate Rouch left to focus on breast cancer recovery; Barret Zoph departed again after returning to OpenAI about five months earlier, with the company not disclosing specific reasons; ethics lead Chloé Bakalar resigned after less than a year; head of security systems Johannes Heidecke left during a restructuring of the security team; and Joshua Achiam, who had worked at OpenAI for nearly nine years, departed after his alignment team was dissolved and he was reassigned to the role of “Chief Futurist.”

The most prominent figure was Caitlin Kalinowski, who led robotics and consumer hardware and resigned in March over concerns regarding OpenAI’s governance of its collaboration with the Pentagon. She publicly emphasized that the core issue was not opposition to government partnerships per se, but rather the lack of adequate governance discussions around red lines such as domestic surveillance and lethal autonomous weapons. Shortly after Dresser’s departure, Kaylin Voss, Vice President of Sales for the Americas, also resigned. Most recently, Chris Malone left his position as head of data centers following OpenAI’s restructuring of its infrastructure management framework.

So, if you're looking for a common reason among these 14 people, the answer is actually: there isn't one.

Some people start businesses after accumulating sufficient wealth; others can no longer sustain high-intensity work due to health issues; some have seen their business units deliberately scaled back by their companies; others have lost their original roles due to organizational restructuring; and some genuinely disagree with the governance approach. Labeling all of these individuals as “bearish on OpenAI” flattens a complex reality into a simplistic, easily spread story that doesn’t hold up under scrutiny.

Conversely, framing this wave of change as “normal Silicon Valley turnover” is equally dismissive. For a company preparing for an IPO with a potential valuation in the trillions, investors care far less about whether the departure stories are dramatic, and far more about whether these departures have strategically targeted critical control points.

And the problem with OpenAI lies precisely here.

If the 14 people who left were all from peripheral experimental projects, the market could interpret this as a resource optimization. But this list is not like that.

On the revenue side, Dresser departed; on the sales side, Voss left and Zoph departed again; on the product and applications side, Simo, Weil, Peebles, and Narayanan successively exited; on the brand side, Rouch stepped down; on the security and ethics side, Bakalar, Heidecke, and Achiam left; on the hardware side, Kalinowski resigned; on the infrastructure side, Malone departed; along with former COO Lightcap.

This essentially covers all the key pillars of a large tech company: how to make money, who to sell to, what products to offer, how to manage risk, and how to scale computing power.

Why is this particularly sensitive before an IPO? Because the core narrative of a private company is growth, while the primary requirement for a public company is growth plus predictability. The secondary market can tolerate a high-growth company operating at a loss temporarily, investing heavily in infrastructure, or even allowing founders to maintain strong control—but it struggles to tolerate prolonged turnover in key roles, constantly shifting lines of responsibility, and frequent contractions in business direction.

OpenAI itself has now reached a scale where it can no longer be explained away as a startup. On August 13, the company disclosed upon appointing its new Chief Revenue Officer that its products now have over one billion weekly active users and more than two million enterprise customers—double the numbers from a year ago. By August, publicly reported figures showed its annualized revenue run rate had surpassed $40 billion, roughly double what it was at the end of 2025.

This is no longer a company that can rely long-term on a few star executives building the structure as they go.

When revenue grows from billions to hundreds of billions, individual heroism can create speed; but when a company must manage millions of enterprise customers, global regulatory relationships, and massive computational capital commitments, the truly scarce capability becomes institutionalized replication: who is responsible, who reviews, who is accountable, how long vacancies remain open, and who ensures continuity after strategic adjustments.

From this perspective, what’s most concerning about the departure of 14 people isn’t the reduction in talent, but that OpenAI is still operating at the pace of a fast-moving startup while managing the commercial scale and societal impact of a quasi-public company.

It must be acknowledged that OpenAI is not passively taking hits.

This year’s personnel changes and business adjustments clearly show signs of deliberate restructuring. Lightcap transitioned from COO to special projects, with commercial responsibilities redistributed; Sora was shut down, and the leader of OpenAI for Science was replaced; the security team was reconsolidated; and infrastructure management underwent a leadership reshuffle. In April’s personnel changes, Greg Brockman also took over the product organization. Taken together, these changes indicate that OpenAI is redefining its power boundaries and reallocating resources toward more core functions directly tied to commercialization and computing power.

The logic behind this is not hard to understand. OpenAI already has no shortage of grand vision; what it lacks is the ability to convert its massive user base, model capabilities, and capital investment into sustainable cash flow. Prior to going public, streamlining non-core initiatives, concentrating resources, and strengthening enterprise business and computing infrastructure makes sense from a corporate governance perspective.

In fact, if management determines that the expansion over the past two years has been too scattered, the closer the company gets to its IPO, the more urgently it should take decisive action rather than entering the public market burdened by historical liabilities.

The problem is that organizational restructuring is never free.

First, frequently changing leadership creates execution friction. New leaders must reconfirm budgets, team boundaries, reporting lines, and priorities; what appears externally as a simple role change may internally result in prolonged decision-making delays.

Second, the re-centralization of power increases "key person risk." OpenAI is no longer the small startup team it was in 2022. If an increasing number of critical decisions ultimately rest with just a few founding-level individuals, decision-making may become faster, but governance discounts may also rise. Public markets reward efficiency but also price in excessive centralization.

Third, the wealth dynamics within OpenAI have shifted. According to The Wall Street Journal this year, more than 600 current and former employees collectively cashed out approximately $6.6 billion through share sales, with some individuals selling up to $30 million in shares. While cashing out wealth is not inherently negative, it changes the logic of retention: when a group of early employees have achieved financial freedom, the appeal of staying must increasingly rely on mission, authority, research freedom, and organizational trust—rather than continuing to depend primarily on the promise of stock options tied to an eventual IPO.

This is precisely why OpenAI’s next true challenge is not whether it can still hire top talent. With its brand, capital, and model resources, hiring has never been the hardest issue. The real challenge is whether the new leadership can establish stable roles and responsibilities, and whether external investors can be convinced that this structure won’t be overhauled again in six months.

In June, OpenAI secretly filed its IPO documents with the U.S. Securities and Exchange Commission. At the time, the company emphasized that it had not yet decided on a timeline for going public and clearly stated that certain things are easier to accomplish while privately held. Increasingly public information now points to a formal listing window around 2027, while market discussions still target a valuation in the trillion-dollar range.

This makes the personnel shifts this year even more significant.

If OpenAI goes public soon, the departure of its 14 key executives would be immediately interpreted as a stability issue ahead of the roadshow; if it doesn’t go public until 2027, then now is truly the final window to fundamentally restructure the organization without immediately facing quarterly earnings scrutiny.

In other words, this round of "major overhaul" could represent either risk exposure or risk cleanup. Both interpretations are currently valid, and the key lies in what happens over the next six to twelve months.

If Dali Rajic can quickly take over the company’s revenue system, the security team stops losing senior leaders after restructuring, the infrastructure department steadily advances its computing power plan, and OpenAI ceases frequently shutting down or redefining its core product lines, then today’s departure of 14 individuals could, in hindsight, be reinterpreted as an organizational streamlining prior to going public.

But if business, security, research, and infrastructure continue to undergo frequent changes, the market will begin to ask a more difficult question: Is this company proactively upgrading, or has its strategic core never been stable?

This is the harshest truth about IPOs. Private markets are willing to pay a high premium for "potential," but public markets ultimately demand "repeatability." OpenAI has excelled at creating the former over the past few years, but it now must prove it can deliver the latter just as effectively.

If you only look at product discussions, the competition between OpenAI and Anthropic is easily framed as which is stronger, GPT or Claude. But from a capital markets perspective, the contest in 2026 has become a different race: who can prove first that they are a supercompany worthy of long-term holding.

As of the end of July, Anthropic’s annualized revenue run rate exceeded $65 billion, up from $47 billion in May and approximately $9 billion by the end of 2025; following its May funding round, the company’s valuation reached $965 billion. In comparison, public reports from August regarding OpenAI indicated an annualized revenue run rate exceeding $40 billion and a recent publicly disclosed private valuation of $852 billion.

It is important to emphasize that the annualized revenue run rate is not the same as audited full-year revenue; their business models, revenue quality, and cost structures are not directly comparable, and one cannot declare a winner based solely on the figures of $65 billion versus $40 billion. However, the trend itself is striking enough: Anthropic is no longer just the “safety-focused startup” trailing behind OpenAI—it has established significant commercial weight in the enterprise and developer markets.

The same applies at the organizational level. At least based on currently publicly available information, Anthropic has not experienced the same density of departures among core leaders across business, security, product, and infrastructure as OpenAI. This does not mean Anthropic lacks internal conflicts, nor does it imply that its governance is necessarily superior; however, capital markets can only price based on observable information. On the roadshow, the fact that “no one keeps asking why another executive left” is itself an advantage.

This is precisely where OpenAI is currently facing real trouble.

Its technical strength, brand influence, user base, and fundraising capabilities remain substantial, far from reaching a state of "decline due to talent loss." On the contrary, OpenAI is still likely one of the few AI companies globally capable of simultaneously advancing models, consumer products, enterprise services, hardware, and massive-scale computing infrastructure.

But precisely because of this, the market will demand more. A startup valued at tens of billions of dollars can rely on its founders to solve many problems; a company aiming for a public market valuation of nearly a trillion dollars must prove that the system will continue to function even if a few key individuals leave.

Therefore, the departure of at least 14 senior leaders and key personnel from OpenAI this year should neither be exaggerated as "the company is falling apart," nor dismissed lightly as "normal staff turnover."

A more accurate assessment is that OpenAI is undergoing a pre-IPO organizational restructuring, and the scale of this adjustment is significant enough to influence how capital markets evaluate its governance.

Its largest asset is still its迅猛 growth: over one billion weekly active users, more than two million enterprise customers, and an annualized revenue run rate exceeding $40 billion—these figures indicate that OpenAI still possesses strong commercial momentum. Its greatest risk, however, is that the organization has yet to demonstrate stability commensurate with this scale: senior leadership positions continue to change, business boundaries are constantly shrinking and being redrawn, security and governance controversies remain unresolved, and infrastructure is entering a more capital-intensive phase.

These two forces are simultaneously pulling on OpenAI.

If growth ultimately overcomes organizational friction, today’s wave of departures will become a brief footnote in an IPO prospectus; if organizational friction begins to undermine execution efficiency, the market will realize that OpenAI’s true cost may not just be compute power, but the complexity of its management itself.

Before the IPO, what Sam Altman most needed to prove may no longer be whether OpenAI can still develop the next, more powerful model—something the market assumes it has the potential to achieve.

The more critical question is: When a company becomes large enough to influence the global AI industry, capital expenditures, and regulatory direction, can it consistently manage technology, business, security, and infrastructure simultaneously without relying on constant restructuring and the overburdened efforts of a few key individuals?

The departure of 14 people will not directly determine OpenAI's fate.

But they have already brought this issue to the attention of all future shareholders.

1. Business Insider, “13 Executives Who Left OpenAI in 2026,” August 21, 2026.
2. The Wall Street Journal, “OpenAI’s Head of Data Centers Has Left the Company,” 2026-08-26.
3. OpenAI, “OpenAI appoints Dali Rajic as Chief Revenue Officer,” 2026-08-13.
4. The Associated Press, “OpenAI files confidential SEC paperwork for IPO, opening the door to a Wall Street debut,” 2026-06-08.
5. Bloomberg, “OpenAI’s Revenue Run Rate Exceeds $40 Billion Ahead of IPO,” 2026-08-13.
6. Reuters, “Anthropic Revenue Run Rate Tops $65 Billion, Source Says,” 2026-08-17.
7. TechCrunch, “OpenAI executive shuffle includes new role for COO Brad Lightcap to lead ‘special projects’,” 2026-04-03.
8. The Wall Street Journal, “How a Job at OpenAI Became the Greatest Lottery Ticket of the AI Boom,” 2026-05.