“Arm's market share in Windows — I think, truly, in the next five years, it could be better than 50%.”
Deadline is mid-2029; as of 2026 Mercury Research estimates Arm-based client CPUs (mostly Apple Macs and Chromebooks) at ~14% of the total client market, with Windows-on-Arm share in the low single digits.
“Demand for AI infrastructure remains strong, with chips, data centers, energy and workers all limiting supply — the industry could stay supply-constrained for the next two to three years.”
Horizon runs to mid-2029; Haas repeated in September 2026 that demand is 'off the charts' and supply constraints could persist for years.
“By the end of the decade, AI data centers could consume as much as 20% to 25% of U.S. power requirements. Today that's probably 4% or less.”
Target year is 2030; mainstream forecasts (LBNL, EPRI) currently put total US data center consumption at roughly 9%–17% of US electricity by 2030, well below Haas's 20–25% figure.
“We project that by the end of the year, Arm's market share in data centers will be 50% — up from about 15% in 2024 — driven by the AI buildout.”
By Arm's own (narrower) metric the target was hit: in its FY2026 results (May 2026) Arm said Arm-based CPUs account for roughly 50% of compute among the top hyperscalers. But on broader third-party measures Arm fell far short — Mercury Research put Arm-based server CPUs at only ~12.5% of total server CPU shipments in Q4 2025.
