Meta Poaches AWS Veteran Dave Brown to Build Meta Compute

The interesting part of the Dave Brown move is not that a senior AWS executive is leaving after nearly two decades. That happens. The interesting part is where he is landing and what he has reportedly been asked to build.

Brown is the AWS senior vice president who ran compute and machine learning services, a group whose roots reach back to the earliest days of EC2. He joined Amazon in 2007 in Cape Town, South Africa, where part of the original EC2 engineering sat, and became part of Andy Jassy's S-team, the roughly 28-person group of top executives who meet with the CEO on the big business calls. According to the Wall Street Journal, he leaves at the end of July 2026 for Meta's infrastructure organization with two jobs: continue the massive data center build-out, and construct something the company is calling Meta Compute.

Meta Compute is the phrase worth staring at. As reported, it looks like a potential pivot toward renting AI infrastructure to outside customers the way AWS, Microsoft Azure, and Google Cloud already do, either by leasing raw compute or by selling access to models running on Meta's hardware. That would make Meta a fourth genuine hyperscaler and turn projects like its Louisiana campus, now expanding from 2 gigawatts to 5 gigawatts with total investment exceeding $50 billion, into revenue rather than sunk cost.

Amazon is not standing still. AWS CEO Matt Garman named Dave Treadwell, a 27-year Microsoft veteran who has led Amazon's eCommerce Foundation since 2016, to take over compute and machine learning services on August 1, according to CNBC. AMZN shares climbed on the reshuffle, which is not what you would expect if the market thought Meta was about to eat AWS's lunch tomorrow.

The honest caveat is that hiring one executive is not the same as running a cloud. None of the reporting pins down what Meta Compute will actually sell, to whom, or on what timeline, and a public cloud needs enterprise sales, quotas, support and account management that Meta has never done at that scale. What the reporting does tell us is that the company already spending tens of billions on AI data centers has poached the person who knows how EC2 grew from an experiment into the backbone of most of the internet. If that build succeeds even partially, GPU-hour pricing at the current three hyperscalers is the first thing that gets uncomfortable.