A brief hot mic moment at a White House dinner sent shockwaves through tech and policy circles after Meta CEO Mark Zuckerberg, pressed by President Trump about investment plans, volunteered that the company expects to spend “around” $600 billion in the United States through 2028. Moments later, still on a live microphone, he leaned back to acknowledge he hadn’t been prepared for the question and wasn’t certain which figure the president wanted. The candid aside fueled instant debate over whether such an eye-popping number is realistic—or strategic.
A candid moment, then a clarification
After the footage circulated, Zuckerberg later signaled on Threads that the $600 billion figure is indeed the plan—and could rise if AI development continues to accelerate. The clarification matters: it shifts the narrative from a flub to a public stake in the ground about Meta’s domestic buildout of AI infrastructure, data centers, and talent. In other words, the hot mic didn’t just pick up a number; it captured a strategic commitment.
Can Meta really spend $600B by 2028?
Meta’s own guidance highlights the scale of the challenge. On its most recent earnings call, the company projected full-year 2025 total expenses of roughly $114–$118 billion, with infrastructure and compensation as key drivers. CFO Susan Li also noted budgeting beyond that window was not finalized, and expenses were expected to grow about 20%–24% year over year. Even if “investment” spans both capital expenditures and operating costs, pushing aggregate U.S. spending to $600 billion in just a few years would require an aggressive ramp from today’s run rate.
Consider the capital intensity. Building AI capacity at hyperscale entails land acquisition, multi-building campuses, advanced cooling, fiber, and tens of thousands of accelerated computing units. Industry analysts routinely peg the cost of a single new hyperscale campus in the billions. Meanwhile, chip availability and power constraints introduce their own ceilings, regardless of budget. The math doesn’t make $600 billion impossible, but it would put Meta on a historically steep spending trajectory.
The AI spending race among tech giants
The dinner showcased a broader arms race. According to accounts from attendees, Google’s Sundar Pichai said the company’s AI outlays were already well over $100 billion and could reach $250 billion within two years. Microsoft CEO Satya Nadella pointed to a near-$80 billion annual U.S. spend. These figures, while not strictly comparable across companies, underscore a clear trend: hyperscalers are pivoting from incremental capex to multi-hundred-billion-dollar commitments to keep pace in AI compute, model training, and global distribution of services.
For context, these pledges exceed the scale of many national infrastructure programs. They also come with a regulatory dimension: as these companies concentrate compute, data, and distribution, antitrust scrutiny, AI safety guidelines, and labor considerations will remain in the foreground. The hot mic provided a snapshot of how political theater and corporate strategy increasingly interlock in the AI era.
Power, permits, and public pushback
Even with cash in hand, electricity and permitting remain bottlenecks. Bloomberg reported that President Trump told the CEOs the administration would streamline approvals for electrical capacity and related infrastructure. Yet, as The Washington Post has documented, communities nationwide are increasingly mobilizing against rapid data center expansion near residential areas, citing noise, land use, and strain on local utilities.
Energy costs compound the challenge. Residential electricity bills are climbing across the Northeast, with some states facing average increases of about 20%, and Virginia—the country’s densest data center hub—sits at the center of the debate. Building enough reliable, low-carbon power for AI campuses requires long-lead generation, transmission upgrades, and grid interconnections that can take years. Money alone won’t compress those timelines.
Why this moment matters
Hot mics are often dismissed as political curios, but this one surfaced a substantive signal: Meta is committing to an audacious U.S. investment plan at a time when AI economics are being rewritten in real time. If Meta executes, the result would be a nationwide buildout of data centers, chip clusters, and high-skill jobs—alongside intensifying questions about energy, land use, and competition.
The stakes go beyond bragging rights. AI leaders are converging on similar conclusions: compute wins markets, and control of power and permitting will decide where that compute can live. The Hill reported that not every tech titan received an invitation to the dinner, but the broader message was unmistakable. Washington wants the investments, the companies want the green lights, and communities want guardrails. Zuckerberg’s unscripted moment put a number on that bargain—and set a high bar for delivery.
Sources: Meta earnings call commentary; statements by attendees cited by Bloomberg and The Hill; community and energy reporting from The Washington Post.