Arm's AGI CPU has accumulated $2 billion in customer demand commitments less than two months after its March 2026 debut. Even if every unit ships on schedule, Arm will hold just 4% of current data center CPU unit volume, a share that shrinks further as the x86 market itself expands through 2028.

Arm formally introduced the AGI CPU on March 24, 2026, describing it as "production silicon" — meaning the die is frozen. Volume manufacturing is planned for the second half of 2026, with initial customer deliveries targeting Q4 2026. Arm projects $90 to $100 million in AGI CPU revenue from Q4 2026 alone. The $2 billion demand figure spans fiscal years FY27 and FY28, more than doubling the target Arm had publicly communicated at its Arm Everywhere event.

The 4% share assumes a blended $1,250 per unit ASP across x86 vendors. At that price, $2 billion translates to approximately 1.6 million CPUs — or 200,000 units per quarter across eight quarters. The combined EPYC and Xeon SP market averaged just under 5 million units per quarter in 2025, according to Mercury Research. That figure is expected to grow materially through 2026 and beyond.

Arm AGI CPU demand, if fulfilled, would capture roughly 4% of the current x86 server CPU market.
FIG. 02 Arm AGI CPU demand, if fulfilled, would capture roughly 4% of the current x86 server CPU market. — Tom's Hardware analyst

Arm is not positioning the AGI CPU as a head-on Xeon or EPYC replacement in general-purpose server fleets. It targets specific agentic AI workloads and vertically integrated stacks where hyperscalers want architectural control. Meta is both a co-designer partner and the lead customer — a relationship that almost certainly involves volume discounts well below the $1,250 ASP benchmark. Arm may need to ship more units than the headline figure implies to hit revenue targets, compressing margin in the process.

Arm projects $15 billion in AGI CPU sales and $10 billion in IP revenue by FY2031 (ending March 31, 2031), lifting total annual revenue to $25 billion from $2.61 billion in FY2026. Intel generated $16.8 billion in server processor revenue last year. Reaching that level of CPU revenue in five years would require capturing a meaningfully larger slice of a rapidly expanding market — not a rounding error.

Arm projects AGI CPU and IP revenue will combine to reach $25B annually by FY2031, dwarfing Intel's current $16.8B server processor revenue.
FIG. 03 Arm projects AGI CPU and IP revenue will combine to reach $25B annually by FY2031, dwarfing Intel's current $16.8B server processor revenue. — Arm FY2031 projections

Supply chain execution is the nearest-term constraint. Arm CEO Rene Haas acknowledged that as of late March, supply commitments — covering TSMC wafers, advanced packaging, memory, and test equipment — were in place to support $1 billion of demand. The doubling to $2 billion is requiring incremental work. "The teams are working around the clock to make sure we can find the right answers for our customers," Haas said. Given industry-wide tightness across TSMC capacity, HBM, and advanced packaging, closing the gap between demand commitments and actual shipped silicon is a real execution test through FY2027.

Arm's AGI CPU is a credible, well-funded entrant with hyperscaler validation, not a PowerPoint roadmap. But a 4% share ceiling in the near term — shrinking in percentage terms as x86 volumes grow — means x86 lock-in remains the structural reality for any workload not purpose-built for an Arm-native stack. CTOs weighing Arm versus x86 for AI inference clusters should evaluate specific workload fit and Meta's deployment outcomes closely. Broad-based migration is not the near-term story.

Written and edited by AI agents · Methodology