Unnamed billionaire tech CEO announces $25 billion backlog of committed AI infrastructure capex, stating 'demand is booked' and comparing buildout scale to 'Great Wall of China.'
During a recent appearance on the All-In Podcast, Cerebras CEO and co-founder Andrew Feldman described an AI infrastructure buildout of unprecedented scale, where compute suppliers are racing to catch up with orders placed months ago. When podcast host Chamath framed the buildout by saying "We've never seen a buildout like this since the Great Wall of China," Feldman pointed to a critical distinction: the industry isn't building on speculation. Much of the demand is already under contract.
"They're not chasing sort of, if you build it, they will come. They're chasing the demand that is booked," Feldman said. Cerebras alone carries a $25 billion backlog, with compute supply unable to keep pace with existing orders from OpenAI, Anthropic, Google, Microsoft, and AWS. The result is a global infrastructure push spanning the US, Europe, the Middle East, and Central Asia—Kazakhstan, Tajikistan, Armenia, and Georgia—where individual data centers consume more power than mid-sized cities.
The financial data from infrastructure providers confirms this booked-demand thesis. NVIDIA posted Q1 FY2027 revenue of $81.615 billion, up 85.23% year over year, with Data Center revenue of $75.246 billion and Data Center Networking growth of 199%. The company guided Q2 revenue at $91.0 billion and disclosed total supply-related commitments of $119.0 billion to serve demand "beyond the next several quarters." CEO Jensen Huang called the expansion "the largest infrastructure expansion in human history."
AMD reported Q1 2026 revenue of $10.25 billion, with Data Center revenue up 57% to $5.78 billion. CEO Lisa Su cited "leading customer forecasts exceeding our initial expectations" on the MI450 Series and Helios platforms, while the company announced a Meta partnership to deploy up to 6 gigawatts of AMD Instinct GPUs. AMD's stock has surged 155.29% year-to-date, the clearest market validation of Feldman's supply-tightness thesis.
Equinix closed 2025 with record annualized gross bookings of $474 million, up 42% year over year, with roughly 60% of the largest Q4 deals driven by AI workloads. The company flagged 52 major expansion projects and added approximately 1 gigawatt to powered land-under-control. CEO Adaire Fox-Martin stated, "Demand for our solutions has never been higher."
Digital Realty Trust signed a 200-megawatt AI inference lease in Q1 2026—the largest hyperscale lease in company history—contributing to $707 million in annualized GAAP base rent bookings. The company holds roughly 1.2 gigawatts under construction and 6.3 gigawatts of buildable capacity in its pipeline.
Feldman acknowledged the debate around whether spending is creating real value, conceding some wasteful spending exists in what he compared to the early experimentation phase of AWS adoption. He likened current AI token usage to shoppers wandering every aisle at Costco before learning to shop efficiently, arguing that net value creation will be enormous and consumption patterns will rationalize over time without undermining the buildout thesis.
The next hard ceiling on AI infrastructure is electricity. Individual data center power loads have doubled from approximately 150 megawatts to 300 megawatts in a single year. The EIA estimates server electricity consumption could reach 818 billion kilowatt-hours by 2050 under a high-demand scenario. If Feldman's thesis holds—that yesterday's orders already exceed today's available compute—the greatest operating leverage will belong to the companies that can pour concrete, secure power, and ship silicon fastest.