Rene Haas has a problem most CEOs would envy. His company, Arm Holdings, designs the processor architecture that powers virtually every smartphone on Earth, most of the world’s tablets, and an accelerating share of data center servers and personal computers. Arm doesn’t manufacture chips. It licenses its blueprints to the companies that do — Apple, Qualcomm, Samsung, MediaTek, Nvidia, and dozens of others. It’s a toll-booth business model, elegant and enormously profitable. And now Haas wants to blow it up.
Not entirely. But enough to make his biggest customers very nervous.
Arm is designing its own CPU — a complete, ready-to-manufacture processor that the company plans to offer directly to device makers and cloud providers. The chip, reportedly in development for months and potentially arriving as early as 2025 or 2026, would put Arm in direct competition with the very licensees who pay it billions of dollars a year. As Wired reported, Haas insists the market needs this product. He may be right. But the strategic risks are enormous, and the fallout could reshape the semiconductor industry’s most important licensing relationship.
The logic behind the move isn’t hard to follow. Arm has watched its architecture conquer mobile computing over the past two decades, and more recently begin displacing x86 processors from Intel and AMD in laptops and data centers. Amazon’s Graviton chips, Apple’s M-series processors, and Microsoft’s push for Arm-based Windows PCs have all validated the architecture’s performance-per-watt advantages. Yet Arm captures only a fraction of the value its technology creates. A licensing fee and a per-unit royalty on every chip shipped — that’s the deal. As chips get more complex and more valuable, Arm has been pushing to increase its royalty rates, a campaign that has already generated friction with licensees.
A complete Arm-designed CPU would change the economics dramatically. Instead of collecting a few percentage points on someone else’s chip, Arm could sell an entire design — or even a finished product — and capture far more revenue per unit. For customers who lack the engineering resources to design their own custom Arm cores, an off-the-shelf option from the architecture’s creator could be genuinely appealing.
That’s Haas’s pitch, anyway.
“There are a lot of companies that want to have an Arm-based chip but don’t have the capabilities to design one,” Haas told Wired. He frames the initiative as serving an underserved segment — companies that want the performance and efficiency benefits of Arm’s architecture but can’t afford the hundreds of millions of dollars and years of engineering effort required to build a custom chip from scratch. Think mid-tier cloud providers, enterprise server buyers, telecom equipment makers, and automotive companies racing to add compute power to vehicles.
It’s a reasonable argument. And it sidesteps the more uncomfortable truth: Arm also wants a bigger piece of the pie that its largest licensees currently dominate.
The semiconductor industry runs on trust. Specifically, it runs on the trust that your IP licensor won’t become your competitor. Arm has spent decades cultivating relationships with chipmakers by promising neutrality — we design the architecture, you build the chips, everybody wins. That neutrality is Arm’s most valuable intangible asset, arguably more important than any individual processor design. The moment Arm ships its own CPU, that neutrality evaporates.
Consider the position of Qualcomm, which pays Arm substantial licensing fees for the architecture underlying its Snapdragon processors. Qualcomm has already been locked in a bitter legal dispute with Arm over licensing terms, a fight that went to trial in late 2024 and resulted in a mixed verdict. The two companies have clashed over whether Qualcomm’s acquisition of chip startup Nuvia — and its use of Nuvia’s custom Arm-based designs — was covered under existing license agreements. Arm argued it wasn’t. A jury sided partly with Qualcomm but left key questions unresolved. The relationship is, to put it mildly, strained.
Now imagine telling Qualcomm that Arm plans to sell a competing chip to the same device manufacturers Qualcomm serves. The diplomatic way to describe Qualcomm’s likely reaction doesn’t exist.
MediaTek, which supplies Arm-based processors for a huge share of the world’s Android phones, faces a similar dynamic. So does Samsung’s semiconductor division. And Nvidia, which uses Arm architecture in its Grace CPU for data centers, might reasonably wonder whether its licensor is about to become a rival in the server chip market it’s aggressively targeting.
Apple is something of a special case. Its custom Arm-based silicon — the M-series for Macs and the A-series for iPhones — is designed entirely in-house, and Apple has an architectural license that gives it broad freedom to modify Arm’s instruction set. Apple is unlikely to buy an off-the-shelf Arm CPU. But Apple is also Arm’s single largest royalty payer, and anything that destabilizes the broader Arm licensing model could eventually affect that relationship too.
The Intel parallel is instructive and cautionary. For years, Intel operated both as a chip designer and a manufacturer, selling its own processors while also offering foundry services to other companies. Customers were perpetually wary of sharing their designs with a company that competed against them. Intel’s foundry business struggled to attract major clients in part because of this conflict of interest. Pat Gelsinger’s strategy to spin out Intel Foundry as a more independent entity was explicitly designed to address this trust deficit. Arm is now walking toward a version of the same problem, except from the opposite direction — moving from pure IP licensing into product sales.
There’s another historical echo worth considering. In the 1990s and 2000s, Microsoft maintained a careful balance as a platform provider, licensing Windows to PC manufacturers while mostly avoiding building its own hardware. When Microsoft finally launched the Surface line of tablets and laptops in 2012, it sent shockwaves through its OEM partner base. Dell, HP, Lenovo — none of them were thrilled to discover that their operating system vendor was now also their competitor. The Surface business eventually found its niche, but the trust damage was real and lasting. Some OEMs diversified toward Chrome OS partly as a hedge.
Arm’s situation carries higher stakes. Microsoft’s OEMs had limited alternatives to Windows at the time. Arm’s licensees, by contrast, have options. They could invest more heavily in RISC-V, the open-source instruction set architecture that has been gaining traction in embedded systems, IoT devices, and increasingly in more powerful computing applications. RISC-V doesn’t require licensing fees. It doesn’t come with a corporate parent that might compete with you. And while RISC-V’s software support and performance still lag behind Arm in most categories, the gap is narrowing.
If Arm’s move into chip design accelerates RISC-V adoption among even a handful of major players, the long-term consequences for Arm could dwarf whatever revenue the new CPU generates. That’s the nightmare scenario for Arm’s investors, who have bid the stock up to extraordinary valuations since its September 2023 IPO. As of mid-2025, Arm trades at multiples that price in years of aggressive growth — growth that depends on maintaining and expanding its licensing relationships, not undermining them.
SoftBank, which owns a roughly 90% stake in Arm, has been pushing the company toward higher-margin businesses and faster revenue growth. Masayoshi Son’s vision for Arm extends well beyond licensing — he sees the company as a central player in the AI computing boom, with Arm’s energy-efficient architecture powering everything from edge AI devices to massive data center clusters. An Arm-branded CPU for AI workloads would fit neatly into that vision. Whether it fits neatly into reality is another question.
The technical execution challenges are substantial. Designing a competitive server or PC processor requires not just a good CPU core — which Arm obviously knows how to build — but also memory controllers, I/O interfaces, interconnects, security features, and extensive validation and testing. It requires building relationships with TSMC or Samsung Foundry for manufacturing. It requires software optimization, driver support, and customer service infrastructure. These are capabilities Arm has never needed before, and building them from scratch is expensive and slow.
Arm could take a shortcut by acquiring or partnering with an existing chip design firm. But any acquisition large enough to matter would likely trigger antitrust scrutiny, given Arm’s central position in the semiconductor supply chain. Regulators in the U.S., EU, UK, and China have all shown willingness to block or heavily condition deals involving critical chip technology. Nvidia’s attempted $40 billion acquisition of Arm, which collapsed in 2022 under regulatory pressure, demonstrated just how sensitive governments are about consolidation in this space.
So Arm is most likely building this capability internally, which means the first product may be a year or more away from volume production. That gives the company time to manage the messaging — but it also gives competitors and licensees time to prepare countermoves.
Qualcomm, for its part, has been investing in custom CPU core design since acquiring Nuvia in 2021. Its Oryon cores, which debuted in the Snapdragon X Elite for laptops, represent a significant step toward reducing Qualcomm’s dependence on Arm’s standard core designs. If Arm becomes a competitor, Qualcomm’s incentive to develop fully custom cores — or to explore RISC-V for future products — only intensifies.
Nvidia has its own motivations to hedge. Jensen Huang’s company already designs custom Arm-based server CPUs and has the engineering talent to go further. An Arm that competes in data center chips could push Nvidia to accelerate its own CPU efforts or explore alternative architectures for future products.
Even Amazon, which designs Graviton processors for its AWS cloud infrastructure, might reconsider its relationship with Arm if the licensing dynamics shift. Amazon has the resources to invest in RISC-V development, and AWS’s scale means even modest per-chip savings on licensing fees could translate into hundreds of millions of dollars over time.
None of this means Arm’s strategy is doomed. There is genuine demand for high-quality, ready-made Arm-based processors among companies that can’t or won’t invest in custom silicon. The automotive industry alone represents a massive opportunity, as vehicles increasingly require powerful compute platforms for autonomous driving, infotainment, and connectivity. Many automakers and Tier 1 suppliers lack semiconductor design expertise and would welcome a turnkey solution from the architecture’s creator.
The enterprise server market offers similar potential. Not every company can be Amazon or Google, designing custom chips for their own data centers. Smaller cloud providers, managed hosting companies, and enterprises running on-premises infrastructure need off-the-shelf options. Today, that market is dominated by Intel and AMD with their x86 processors. An Arm-based alternative, designed by the company that knows the architecture best, could be genuinely competitive — especially for power-constrained environments where Arm’s efficiency advantages shine.
But Haas has to thread an extraordinarily narrow needle. He needs to convince existing licensees that Arm’s new chip targets customers they don’t serve, not customers they’re already fighting over. He needs to maintain licensing revenue from companies that will inevitably view Arm with greater suspicion. And he needs to deliver a product that’s actually good enough to justify the strategic risk.
That last point matters more than it might seem. If Arm’s first CPU is mediocre — competitive with mid-range offerings but not best-in-class — it will have sacrificed partner trust for marginal revenue. The product needs to be exceptional, or the whole exercise becomes a case study in strategic self-harm.
Wall Street will be watching the royalty rate negotiations closely. Arm has been pushing to shift from per-chip royalties to per-device royalties, a change that would significantly increase its revenue from smartphones and other products that contain multiple Arm-based chips. Licensees have pushed back hard. The existence of a competing Arm-branded chip could give licensees additional leverage in those negotiations — “lower the royalty rates, or we’ll accelerate our RISC-V investments” — or it could make them more desperate to maintain good relations with a company that now holds even more power over their businesses.
The geopolitical dimension adds another layer of complexity. China represents a significant share of Arm’s licensing revenue, and Chinese chipmakers have been among the most enthusiastic adopters of both Arm and RISC-V architectures. U.S. export controls have already complicated Arm’s ability to license advanced technology to Chinese companies. If Arm begins selling complete chips, those products could face additional export restrictions, limiting the addressable market. Meanwhile, Chinese companies might view Arm’s vertical integration as yet another reason to accelerate their investment in domestically controlled alternatives like RISC-V.
Haas is betting that Arm’s brand, its deep understanding of the architecture, and its relationships across the industry will allow it to expand into chip design without destroying the licensing business that funds everything else. It’s a bet that plenty of platform companies have made before — sometimes successfully, as with Apple’s move into silicon, and sometimes disastrously, as with various attempts by software companies to compete with their own partners.
The difference is that Arm occupies a uniquely central position in the global semiconductor supply chain. Its architecture is embedded in billions of devices, and its licensing model has enabled an extraordinarily diverse and competitive chip design industry. Disrupting that model isn’t just a business decision. It’s a structural change to how a critical technology industry operates.
Rene Haas clearly believes the opportunity is worth the risk. His job now is to convince everyone else — partners, investors, regulators, and customers — that he’s right. The semiconductor industry will be watching every move he makes, because the outcome won’t just determine Arm’s future. It will shape the competitive dynamics of chip design for the next decade.
And if he’s wrong, the damage won’t be easy to undo.
WebProNews is an iEntry Publication