Meta infrastructure head Santosh Janardhan steps down amid AI build-out

Santosh Janardhan, Meta’s longtime head of infrastructure and co-head of engineering, is stepping down, Meta confirmed on November 11, 2026. He is being replaced by Dave Brown, a former Amazon executive who joined Meta as vice president of infrastructure earlier this year.

The change comes at a point when Meta’s spending on infrastructure has grown sharply as it tries to keep pace with rivals in AI, and less than a year after Janardhan was put in charge of Meta Compute, the company’s initiative to scale its data center investments and strategy. Janardhan is not leaving immediately: Meta says he will stay on through a transition in an advisory role on Meta Compute, and he is expected to leave next year.

What Meta confirmed about Janardhan’s departure

Bloomberg first reported the departure, citing an internal memo that had recently circulated inside the company. Meta then confirmed the change and the successor to TechCrunch. A Meta representative described the handover this way:

“Santosh will stay on for transition and he is moving into an advisory role on Meta Compute.”

That quote comes from Meta’s statement to TechCrunch. What is still missing is a reason. TechCrunch reports that the memo did not explain why Janardhan is stepping down, and that Meta did not immediately respond to a request for clarification. The memo apparently cited the need to “pass the baton”. No source describes any disagreement, restructuring or performance issue behind the move, and none should be assumed.

Janardhan’s tenure is long by industry standards. He originally joined the company in 2009, and TechCrunch credits him with helping Meta grow into the internet presence it is today. The sources do not say when he took on the head of infrastructure title. The earliest description of him in that role in these sources is a May 2023 Meta post, which calls him Meta’s Head of Infrastructure and says he is responsible for developing and operating the hardware, network, software and data centers that all of Meta’s services run on.

Item	What the sources say
Departing executive	Santosh Janardhan, head of infrastructure and co-head of engineering
Joined Meta	2009
Successor	Dave Brown, vice president of infrastructure since earlier in 2026
Successor’s background	19 years at Amazon, focused on compute, cloud and AI
Janardhan’s interim role	Advisory role on Meta Compute during the transition
Expected exit	Next year
Stated reason	Not specified; memo reportedly cited the need to “pass the baton”

Who Dave Brown is

Brown is not a new face at Meta. He joined earlier in 2026 as vice president of infrastructure, so the succession runs to someone already inside the infrastructure organisation rather than to an outside hire announced alongside the departure.

Before Meta, Brown spent 19 years at Amazon, where his LinkedIn profile shows he focused on compute, cloud and AI, according to TechCrunch. That background matters for the job he is taking on. Meta runs its own data centers for its own products rather than selling cloud capacity, but the problems of building and operating very large compute fleets, procuring chips and planning capacity overlap heavily with the work of a major cloud provider.

The sources do not describe Brown’s plans, any changes to his reporting line, or whether he will also take on Janardhan’s co-head of engineering responsibilities. Those details are open questions for now.

Janardhan’s role in Meta’s AI compute build-out

In 2026 Janardhan became one of the public faces of Meta’s AI infrastructure expansion. In February he announced, in a LinkedIn post, that Meta was deepening its partnership with AMD in a deal that would bring up to 6GW (gigawatts, a measure of electrical power that has become a shorthand for data center scale) of AMD Instinct GPUs into Meta’s data centers.

He described the AMD work as co-design “end-to-end, from the chip to the rack, optimizing for performance per watt and per dollar”. In the same post he set out the company’s approach to chips more broadly: a “flexible, portfolio-based strategy” that combines Meta’s own superclusters, Prometheus and Hyperion, with its MTIA in-house silicon program and outside partners such as AMD. His reasoning, in his own words, was that “no single chip can meet the demands of building personal superintelligence”.

The post also gives a direct view of how Janardhan framed his job. “The future of AI cannot be defined by models alone– it will be defined by the infrastructure that powers them at scale,” he wrote, adding: “Compute is the currency of next-generation AI and we’re at the frontlines leading it”. He wrote that diversifying partners and the technology stack was meant to build “a more resilient infrastructure that can adapt quickly as workloads evolve”.

How Meta Compute was set up in January

Meta Compute is the context for this departure. Janardhan’s February post describes it as a “top-level initiative to ensure Meta can deliver world-class AI to billions of people,” launched the month before.

When Meta set it up in January, CEO Mark Zuckerberg laid out the scale of the ambition in a Facebook post reported by PYMNTS:

“Meta is planning to build tens of gigawatts this decade, and hundreds of gigawatts or more over time. How we engineer, invest, and partner to build this infrastructure will become a strategic advantage.”

That quote is from Zuckerberg’s post as reported by PYMNTS. At the time, Meta promoted two executives to run Meta Compute. Janardhan was to keep overseeing the company’s fleet of data centers and their technical architecture. Daniel Gross, a co-founder of Safe Superintelligence who had joined Meta the year before, was to oversee a new group focused on “long-term capacity strategy, supplier partnerships, industry analysis, planning, and business modeling”.

Alongside those two, Meta appointed Dina Powell McCormick, a former Meta board member and one-time adviser to President Donald Trump, as president and vice chairman, to work on “partnering with governments and sovereigns to build, deploy, invest in, and finance Meta’s infrastructure”. In the same period Meta signed agreements with three nuclear energy companies, Vistra, TerraPower and Oklo, to help power its data centers; the company said these, together with an earlier agreement with Constellation Energy, make it one of the top buyers of nuclear energy in the United States.

Put together, the January structure split Meta Compute into distinct lanes: the physical data center fleet and its architecture under Janardhan, capacity and supplier strategy under Gross, and government and financing partnerships under Powell McCormick. Janardhan’s move to an advisory role means the lane closest to the hardware is the one changing hands.

What it means for developers building on Meta’s AI

For developers, the direct impact is limited and, on the evidence available, unclear. None of the sources mention any change to Meta’s AI products, models, chip partnerships or data center projects as a result of the leadership change. Meta has framed it as a transition, with Janardhan staying on in an advisory capacity on Meta Compute.

The broader point is about how much of AI capability now depends on infrastructure decisions made far from the model layer. Meta’s own leaders describe compute as the constraint and the competitive edge: Zuckerberg called the way Meta builds infrastructure “a strategic advantage”, and Janardhan described compute as “the currency of next-generation AI”. Choices such as how many GPUs come from AMD versus in-house MTIA chips, and how quickly new superclusters come online, shape what models a company can train and serve, and at what cost.

A handover in that role is therefore worth tracking for anyone who relies on Meta’s AI products or models, even if nothing changes in the near term. Some practical points:

• Continuity is the stated plan. Brown is already inside Meta’s infrastructure organisation, and Janardhan is staying through a transition.
• The multi-supplier chip strategy was set under Janardhan. Whether his successor keeps the same balance between in-house silicon and partners is not addressed in any source.
• The scale target has not been withdrawn. No source reports any change to the “tens of gigawatts this decade” plan Zuckerberg described in January.

Open questions

Several things remain unknown from the reporting so far:

• Why now. The memo reportedly did not specify reasons beyond the need to “pass the baton,” and Meta did not immediately respond to TechCrunch’s request for clarity.
• The co-head of engineering role. TechCrunch says Janardhan also served as co-head of engineering. The sources do not say whether Brown takes on that role or whether it changes.
• Meta Compute’s leadership after the transition. Janardhan was tasked with leading Meta Compute earlier this year, and the January announcement placed him alongside Daniel Gross. How responsibilities are divided once Janardhan becomes an adviser has not been described.
• Timing of the final exit. TechCrunch reports he is expected to leave next year but gives no specific date.

What to watch next

The next signals will come from how Meta presents Brown’s role publicly, whether the company clarifies the reason for Janardhan’s departure, and whether any changes follow in Meta Compute’s structure. Developers and infrastructure watchers should also look for any update on Meta’s chip mix, including the AMD deployment and the MTIA program, and on its supercluster projects Prometheus and Hyperion, since those are where a new infrastructure leader’s priorities would first show up.