Arm Holdings is no longer content to be the quiet architect behind the world’s smartphones. The British chip designer, majority-owned by SoftBank, has declared its ambition to build processors capable of powering artificial general intelligence — the kind of machine cognition that can match or exceed human-level reasoning across virtually any task. It’s a staggering claim. And investors are trying to figure out whether it’s visionary or delusional.
At the company’s recent investor day, CEO Rene Haas laid out a roadmap that positions Arm at the center of the AI hardware race, a contest most observers assumed was Nvidia’s to lose. Haas told attendees that Arm is designing chips specifically for AGI workloads, targeting what he described as a $100 billion total addressable market in AI infrastructure by 2030. The stock surged on the presentation, climbing more than 5% in a single session, as Yahoo Finance reported.
But here’s the thing. Arm doesn’t actually manufacture chips. It licenses processor designs to companies like Apple, Qualcomm, Samsung, and dozens of others who then fabricate the silicon themselves. This asset-light model has been Arm’s defining feature since its founding in 1990 — a business built on royalties and licensing fees rather than fabs and factories. What Haas is now proposing is a fundamental expansion of that model into the most computationally demanding frontier in technology.
The AGI chip initiative represents Arm’s attempt to capture a far larger slice of revenue per device. Currently, Arm earns royalties that average a few cents per chip in smartphones. In AI data centers, the company believes it can command royalties that are orders of magnitude higher, potentially several dollars per chip, because the processors are vastly more complex and the end products far more expensive. During the investor presentation, Arm projected that its royalty revenue per chip in AI applications could be 10 to 50 times greater than what it earns in mobile.
That math matters enormously.
If Arm can establish its architecture as a viable alternative — or complement — to Nvidia’s GPU-dominated AI training infrastructure, the revenue implications are transformative. Nvidia currently commands roughly 80% of the AI accelerator market, a dominance built on years of software investment in its CUDA programming platform and an enormous installed base of developers who’ve built their models around Nvidia hardware. Displacing that isn’t simple. It isn’t even clear it’s possible in the near term.
And yet the market dynamics are shifting in ways that favor Arm’s pitch. Hyperscale cloud providers — Amazon, Google, Microsoft, Meta — are all designing custom AI chips using Arm’s architecture. Amazon’s Graviton processors and Google’s Axion chips both run on Arm designs. These companies are motivated by a straightforward desire: reduce their dependence on Nvidia, which has pricing power that makes procurement executives wince. Every dollar these tech giants can save on AI infrastructure drops straight to the bottom line, and Arm is the architecture enabling that diversification.
The timing of Arm’s AGI announcement also coincides with a broader industry reckoning about the sustainability of current AI spending. Capital expenditures on AI infrastructure across the major cloud providers are expected to exceed $200 billion in 2025 alone, according to estimates from multiple Wall Street analysts. That spending has powered Nvidia’s meteoric rise — the company’s data center revenue more than tripled year-over-year in its most recent fiscal year. But there’s growing anxiety that the return on all this investment remains uncertain, and customers are actively seeking ways to bring costs down.
Enter Arm’s proposition. The company argues that its chip architectures are inherently more power-efficient than the x86 designs from Intel and AMD, and that this efficiency advantage becomes critical as AI models scale toward AGI-level complexity. Training a frontier AI model already consumes enough electricity to power a small city for months. The energy costs are becoming a strategic bottleneck. Arm’s pitch is that its designs can deliver more computation per watt, which translates directly into lower operating costs for data center operators.
Not everyone is convinced.
Skeptics point out that Arm’s AI ambitions, while compelling on paper, face enormous execution risk. The company has limited experience in the high-performance computing segment compared to Nvidia, which has spent more than a decade refining its AI hardware and software stack. Nvidia’s CUDA platform has become something close to an industry standard, with millions of developers trained on it and countless AI frameworks optimized for it. Switching costs are real and substantial. A chip designer with better theoretical efficiency still has to convince developers to rewrite code, and that’s a harder sell than any investor presentation can convey.
There’s also the competitive question of AMD, Intel, and a growing roster of AI chip startups — companies like Cerebras, Groq, and SambaNova — all chasing the same opportunity. The AI chip market is attracting more capital and more competitors than at any point in its history. Arm’s licensing model means it doesn’t compete directly with most of these players; rather, it provides the underlying architecture they build upon. But that also means Arm’s success is partially dependent on its licensees winning in the market, a variable the company can influence but not control.
SoftBank’s role in all of this deserves scrutiny. Masayoshi Son’s conglomerate owns approximately 90% of Arm’s outstanding shares, a concentration that gives SoftBank enormous influence over the company’s strategic direction. Son has been vocal about his belief that AGI is coming soon and that it will reshape every industry on earth. His enthusiasm for AI is well documented — and occasionally excessive, as the WeWork debacle demonstrated about his broader investment judgment. But Son’s conviction has also led to some prescient bets, and his willingness to fund Arm’s long-term R&D ambitions gives the company a financial runway that most competitors lack.
During the investor day, Arm also disclosed plans to expand its presence in edge AI — the deployment of artificial intelligence capabilities directly on devices like smartphones, cars, and industrial equipment rather than in centralized data centers. This is arguably where Arm’s existing strengths are most formidable. The company’s chip designs already power virtually every smartphone on the planet, and as AI features become standard in consumer devices, Arm stands to benefit from both higher royalty rates and increased chip complexity. Apple’s latest iPhone processors, which include dedicated AI accelerators, are built on Arm architecture. So are Qualcomm’s Snapdragon chips, which power most Android flagship devices.
The edge AI opportunity is more immediate and less speculative than the AGI play. Analysts at Morgan Stanley have estimated that edge AI could add $2 billion to $4 billion in incremental annual revenue for Arm by 2028, driven by higher royalty rates on AI-capable chips across smartphones, PCs, and automotive applications. That’s meaningful growth for a company that reported $3.2 billion in total revenue for its most recent fiscal year.
But it’s the AGI ambition that has captured Wall Street’s imagination — and inflated Arm’s valuation to levels that demand extraordinary execution. At recent prices, Arm trades at roughly 75 times forward earnings, a multiple that prices in years of aggressive growth. By comparison, Nvidia trades at approximately 35 times forward earnings, and even that valuation assumes continued dominance in AI hardware. Arm’s premium reflects both the optionality of its AGI strategy and the scarcity value of its position as the only publicly traded pure-play on processor architecture licensing.
The valuation question is the sharpest point of debate among institutional investors. Bulls argue that Arm is a toll road on the AI buildout — every chip that uses its architecture generates a royalty, regardless of which company manufactures it or which end market it serves. That diversified exposure to AI spending, they contend, justifies a premium multiple. Bears counter that the stock already reflects a best-case scenario and that any stumble in execution, competitive setback, or slowdown in AI capital expenditure could trigger a painful correction.
Both sides have a point.
What’s undeniable is that Arm occupies a unique position in the semiconductor industry. It doesn’t compete with its customers. It doesn’t bear the capital costs of manufacturing. And its architecture is embedded so deeply in the global technology supply chain that switching away from it would require a wholesale reimagining of how chips are designed. That structural advantage is real, and it provides a foundation for the company’s AI ambitions that few competitors can replicate.
Whether that foundation is strong enough to support a $150 billion market capitalization and an AGI chip strategy that won’t generate meaningful revenue for years — that’s the bet investors are making right now. The semiconductor industry has a long history of rewarding companies that correctly anticipate the next major computing wave. It has an equally long history of punishing those that overreach.
Arm is swinging big. The next few years will determine whether Rene Haas and Masayoshi Son have read the future correctly, or whether the company’s ambitions have outrun the physics of both silicon and markets. For now, the stock price says the market is willing to give them the benefit of the doubt. That confidence, like all things on Wall Street, is borrowed against results that haven’t arrived yet.
WebProNews is an iEntry Publication