What’s Happening at OpenAI with Executive Departures?

Chris Malone left OpenAI last week.

His title was Head of Data Centers. In the summer of 2026, this is probably the most critical position that should not be vacant in an AI lab—over 500 towns across the U.S. are restricting data center construction, Texas is reviewing all grid connection applications, New York has imposed a year-long moratorium, and every AI company is competing for the same batch of power access permits.

Now, the person in charge of this matter has left.

Who is he, and why does his departure stand out?

Malone joined OpenAI in March 2025, shortly after the announcement of the Stargate project—a bold initiative by OpenAI in collaboration with Oracle and SoftBank to build its own data centers.

His background was perfectly suited for this role: before joining OpenAI, he spent nearly five years at Meta as a distinguished engineer, leading the company’s data center strategy; prior to that, he worked for over a decade at Google as a distinguished engineer and senior director, focusing on data center technology.

The problem is, Stargate got off to a rocky start. OpenAI subsequently pivoted to signing contracts with cloud vendors for computing power instead of building its own facilities. Now it is restarting some of its self-built efforts—not just renting chips but leasing entire facilities. However, someone else is leading this effort, not Malone.

Changes in the organizational structure further illustrate the issue. Malone originally reported directly to President Greg Brockman. Earlier this year, the infrastructure department was reorganized, with Vice President Sachin Katti taking over a larger group that reports to Brockman, while Malone and another leader, Adrian Caulfield, became co-leads of a "Data Center Technology Engineering and Design" team.

In July, OpenAI promoted Uday Ruddarraju to Chief Technology Officer for Computing Capacity, reporting directly to Brockman; Brent Mayo, who was poached from Elon Musk’s xAI, reports to Ruddarraju and is responsible for ensuring computing projects are completed on time. Both of these individuals previously worked on Musk’s Colossus supercomputer in Memphis.

This means that before his departure, the weight of this position had already been diluted.

OpenAI’s official response sidestepped this point: "Earlier this year, we reorganized our infrastructure organization to support the scale and pace of our work. We have a strong, highly experienced data center team with clear leadership and the technical expertise needed to execute our plans."

Why are so many people leaving?

Malone is not an isolated case. He is the tenth or so this year.

According to The Wall Street Journal, at least 12 high-profile employees have left, while Business Insider counts 13, several of whom left in the past month. And these are not junior employees; they hold some of the highest positions in the company: Chief Revenue Officer, Chief Operating Officer, Chief Marketing Officer, Chief Product Officer.

The list includes:

Product head Kevin Weil left in April;

Fidji Simo—Sam Altman’s second-in-command, overseeing much of OpenAI’s core business—resigned last month after a medical leave;

Longtime executive and Chief Operating Officer Brad Lightcap announced his departure this month to pursue new projects;

Chief Revenue Officer Denise Dresser, who was only brought in from Slack’s CEO position last December, left after just eight months.

Two weeks ago, OpenAI appointed its second Chief Revenue Officer within a year—Dali Rajic, president and COO from Alphabet’s Wiz.

Typically, startup employees can only cash in their stock after the company is acquired or goes public, which may involve a lock-up period. This is known as "golden handcuffs": you have to stay to get that money.

However, OpenAI gave employees a financial incentive in a funding round last year, allowing them to sell a total of $6.6 billion in shares.

With the money in hand, the handcuffs loosened. Leaving suddenly became much easier.

Why isn’t anyone worried about the departures at Google?

Now, OpenAI is gearing up for an IPO, expected in 2027, with executives hoping for a valuation of over $1 trillion.

At this critical juncture, any perception of "top-level people running away" will make investors uneasy. Even if Brockman is right and the actual operational impact is minimal, perception itself is an asset.

Moreover, talent loss coincides with two other ongoing issues: one is skepticism about the business model, and the other is competitive positioning. In the second quarter, OpenAI’s revenue grew by 18% year-on-year—during the same quarter, competitor Anthropic’s revenue more than doubled. It is now chasing Anthropic in sales to enterprise clients.

On the other hand, the departures at Google are significant as well.

Chief Scientist Jeff Dean left this month after 27 years at the company to start his own venture, taking three senior engineers with him. In June, Nobel laureate John Jumper and AI pioneer Noam Shazeer also departed—Shazeer’s departure is particularly painful, as Alphabet had just spent billions to bring him back two years ago.

But Google has a deep bench. It acquired the UK’s DeepMind 12 years ago, and that division is now the cornerstone of its AI business. It even invested in Dean’s new company.

Replacement is quick: Demis Hassabis moved from DeepMind to take Dean’s position, and Koray Kavukcuoglu took over the Gemini model and AI research—Kavukcuoglu had been there even before Google acquired DeepMind. Analysts believe Hassabis is a visionary thinker, while Kavukcuoglu is more operationally focused, ensuring that profitable AI technologies are delivered on time.

For Google, which has recently fallen behind in the AI race, this turnover might even be healthy.

The difference in departures lies in whether there are people to fill the gaps behind them.

The poachers are their common suppliers.

In the same week, there was another personnel news: NVIDIA secured a $6 billion authorization agreement to take over more than 100 employees from the startup Poolside.

These individuals are expected to join NVIDIA’s Nemotron open-weight model project, with a clear goal—to create a U.S. counterweight to the powerful open-source models recently emerging from China.

This situation has two sides. One side is the national security narrative: models released by Chinese companies like Moonlight and Zhipu have indeed made Washington nervous. The other side is the business relationship: NVIDIA supplies chips to closed-source developers like OpenAI and Anthropic, and now it wants to use its own chips to create a cheaper, competitive open-source model.

And this path is two-way. On the same day Malone’s departure news broke, OpenAI announced the benchmark results of its self-developed chip Jalapeño at the Hot Chips conference, claiming it surpassed NVIDIA’s GB300 in throughput and response speed per kilowatt, achieving up to 1.9 times the throughput and 3.6 times lower latency at 700 watts versus 1400 watts. This chip was developed in collaboration with Broadcom, designed solely for inference, with plans for small-scale deployment by the end of this year and broader rollout in 2027.

Suppliers are moving into the customers’ territory, while customers are encroaching on the suppliers’ territory.

However, there is one thing both sides still need to resolve: those chips ultimately need to be installed in data centers, which need to be connected to power. And on OpenAI’s side, the person in charge of this matter has just left.