ARM Holdings: The AI-Driven Server Play With a Price to Pay

Rhys NorthwoodFriday, May 16, 2025 2:29 pm ET
27min read

The server chip market is undergoing a seismic shift, and ARM Holdings (ARMH) is at the epicenter of it. With a bold 2025 target to claim over 50% of the global data center CPU market—up from 15% in 2024—the company is betting big on AI-driven demand. But with a P/E ratio soaring to 61x, investors face a critical question: Is ARM’s valuation ahead of itself, or is this a once-in-a-decade structural play? Let’s dissect the opportunity and the risks.

The Bull Case: ARM’s Server Market Surge

ARM’s ambitions are underpinned by three unstoppable forces:

  1. AI Workload Dominance:
    Data center operators like AWS, Google, and Microsoft are shifting to ARM’s power-efficient architectures to handle AI training and inference. AWS’s Graviton4 chips now power over 50% of its new server instances, while Google’s Axion is operational in 10 regions and used by Spotify and Siemens. NVIDIA’s Grace CPUs—pairing 144 Arm Neoverse V2 cores—form the backbone of its GB200/GB300 AI servers, now deployed by hyperscalers globally.

  2. Partnerships with Giants:
    Microsoft’s Azure is rolling out Cobalt 100 chips (Arm-based) for enterprise workloads, while NVIDIA’s Grace Blackwell Superchip is set to power Microsoft’s AI infrastructure at scale. The duo’s $80 billion data center investment in 2025 will further cement NVIDIA’s GPU-ARM CPU hybrid systems as the gold standard for AI.

  3. Royalty Uplift:
    ARM’s Armv9 architecture and Compute Subsystems (CSS) are driving record revenue. Full-year 2025 revenues hit $4 billion, with royalties surging past $2 billion for the first time. Data center chips, which incorporate more of ARM’s IP than smartphones, generate 2x higher royalty rates, fueling margins.

The Bear Case: Valuation and Execution Risks

Despite the growth story, three red flags demand scrutiny:

  1. Overvaluation Alert:
    Trading at a P/E of 61x versus the industry’s 26.6x, ARM’s stock is pricing in perfection. Even if it hits its 50% market share target, skepticism lingers about whether investors are overpaying for 2025’s growth today. A 10% earnings miss could trigger a sharp correction.

  2. Regulatory and Legal Headwinds:
    Qualcomm’s ongoing antitrust lawsuit—a $2 billion dispute over licensing—could disrupt partnerships or delay product launches. Meanwhile, U.S.-China trade tensions and tariffs threaten supply chains for global hyperscalers reliant on ARM-based chips.

  3. X86 Entrenched Dominance:
    Intel and AMD’s x86 architecture still holds 85% market share, supported by legacy software ecosystems. While ARM’s ecosystem is growing (22 million developers, 8 billion Kleidi toolchain installs), full software stack parity remains years away.

Strategic Recommendation: Wait for a Pullback

ARM’s long-term thesis is undeniable—AI infrastructure is the next computing frontier, and ARM’s power efficiency is unmatched. However, the current valuation leaves little room for error. Here’s how to play it:

  1. Target a 20% Correction:
    Wait for a pullback to £50–£55 (from current £63) before averaging into positions. This aligns with a P/E compression to 45x, more in line with its growth peers.

  2. Focus on Execution Milestones:
    Watch for:

  3. NVIDIA’s Grace Blackwell shipments hitting 100,000 units by end-2025.
  4. Microsoft Azure’s Cobalt adoption surpassing 20% of its server fleet.
  5. ARM’s Malaysia AI deal scaling beyond its initial $1 billion commitment.

  6. Hedging Against Risks:
    Pair ARM with NVIDIA (NVDA) or Microsoft (MSFT) to capture upside while mitigating ARM-specific risks. Both are direct beneficiaries of its server chip success.

Conclusion: A Buy at the Right Price

ARM is a decade-defining play in AI infrastructure—but at today’s prices, it’s a high-wire act. The structural tailwinds are undeniable, but the stock’s premium demands patience. For investors willing to wait for a 15–20% retracement, ARM offers asymmetric upside. As the saying goes: "Don’t let perfect be the enemy of good." This is a company to own, not to chase.

Investment Grade: ★★★☆☆ (Hold for a pullback)
Target Entry Range: £50–£55 (Potential 15–20% downside from current levels)
Hold Period: 3–5 years for market share targets to materialize.

The server chip war is just heating up. ARM’s future is bright, but its present price is demanding. Proceed with caution—and wait for the dip.