Arm Holdings CEO Rene Haas Has a Big Warning for Intel and AMD

Arm says its first production data center silicon could cut AI data center capital expenditure by up to $10 billion per gigawatt and deliver more than 2x the performance per rack versus x86 platforms. The company also said it sees more than $2 billion in AGI CPU revenue over the next couple of fiscal years and up to $15 billion in annual revenue by fiscal 2031, though AMD and Intel remain far ahead today with Q1 data center revenues of $5.8 billion and $5.1 billion, respectively. The piece is constructive for Arm’s long-term AI positioning, but it frames the impact on AMD and Intel as limited near term.

Analysis

Arm’s real threat is not immediate revenue displacement, but pricing power erosion across the AI server stack. If hyperscalers can use Arm-based designs to lower rack-level capex and power draw, procurement teams will pressure x86 vendors to concede margin or fund more of the integration burden through custom silicon, software, and packaging rebates. That makes this less about unit share next quarter and more about a multi-year gross margin reset for AMD and Intel in data center CPUs.

The second-order winner may be the hyperscalers themselves, not Arm. Lower capex per gigawatt improves return on AI infrastructure and should accelerate buildouts, which is constructive for the whole compute ecosystem, but especially for firms that monetize workload growth rather than chip ASPs. In that sense, the market may be underestimating the positive read-through for META, AMZN, and MSFT if cheaper inference economics pull more AI applications into production faster than consensus models assume.

The contrarian angle is that Arm’s addressable revenue path is still small relative to the incumbent installed base, so the stock reaction risk is asymmetrical over the next 1-2 quarters. The market could overreact to headline share-shift estimates while missing the slower adoption bottlenecks: software porting, validation cycles, and enterprise procurement conservatism. If Arm’s deployment ramp slips by even 1-2 quarters, the competitive narrative cools quickly, but if it lands with Meta and one more hyperscaler, the x86 multiple compression becomes a 12-24 month story rather than a near-term trade.

Key Decisions for Investors

• Reduce AMD vs. NVDA exposure into strength over the next 2-4 weeks; AMD is most exposed to any market share/premium-margin debate in data center CPUs, while NVDA retains the stronger platform lock-in and AI software moat.
• Short INTC calls or use put spreads 3-6 months out if the stock continues to price in an uninterrupted AI inference tailwind; downside risk is less about revenue collapse than multiple compression if Arm adoption accelerates faster than expected.
• Long META / short INTC or AMD as a pair trade for 6-12 months: META benefits from lower AI infrastructure cost per unit of compute, while the short leg captures pricing pressure on x86 vendors. Favor this if capex guidance remains elevated but inferencing efficiency keeps improving.
• Buy out-of-the-money AMD downside puts into any post-earnings rally; the setup is attractive if consensus extrapolates current server growth too far ahead of actual Arm conversion timelines.
• Avoid chasing ARM after the initial enthusiasm unless you can underwrite a 12-24 month adoption curve; the cleaner expression is via hyperscaler beneficiaries rather than the stock most exposed to execution risk.