Arm revenue hits record $1.44bn, but is its $2bn chip pivot changing the risk?

Arm Holdings plc (Nasdaq: ARM) reported record fiscal second-quarter 2027 revenue of $1.44 billion, up 27% year over year and ahead of Wall Street expectations, as artificial intelligence infrastructure continued driving demand for its processor architecture. License and other revenue increased 50% to $773 million, while data-cententre royalties doubled and adjusted earnings per share rose 33% to $0.52. Yet Arm shares remained under pressure around the results, with the stock falling roughly 5% during the regular November 4 session amid broader semiconductor weakness and slipping again during the earnings call.

The reaction reflects an increasingly complicated Arm investment thesis. The company is gaining share inside hyperscale data centres, licensing revenue is accelerating and Chief Executive Officer Rene Haas expects Arm-based server architecture to exceed 60% of deployments at major cloud providers by the end of 2026. At the same time, Arm is moving beyond its traditional intellectual-property model and expects as much as $2 billion of revenue from its own AGI CPUs across fiscal 2027 and fiscal 2028.

That strategic shift could dramatically enlarge Arm’s addressable market, but it changes the economics investors have historically valued. Arm’s licensing and royalty model produces gross margins close to 100% because customers bear much of the manufacturing and inventory risk. Selling complete processors exposes Arm more directly to semiconductor supply, memory availability, working capital, customer concentration and product-cycle execution.

Why did Arm’s licensing revenue surge while royalty revenue missed expectations?

License and other revenue reached $773 million, up 50% from $515 million a year earlier and well above the roughly $625 million analysts had expected. Royalty revenue rose only 8% to $670 million, below the approximately $721 million Wall Street estimate. The divergence matters because licensing and royalties represent different stages of Arm’s economic model.

Licensing revenue is generally generated when customers gain access to Arm architecture, core designs or broader technology packages for future chips. Royalty revenue arrives later when those products move into production and customers ship semiconductor units containing Arm technology. Strong licensing can therefore be interpreted as a leading indicator of future royalties, but the timing between a licence agreement and large-scale chip shipments can extend across several years.

Arm said annualised contract value increased 12%, while large customers continue adopting newer Arm Total Access and Compute Subsystems arrangements. Management also cautioned that licence revenue can fluctuate materially from quarter to quarter because large contracts and accounting treatment can create significant timing effects. Investors consequently should not assume $773 million is automatically the new quarterly base.

The royalty miss is therefore more important than the headline revenue beat might initially suggest. Arm still needs the wave of newly licensed designs to convert into shipped chips before the current licensing boom becomes a durable recurring royalty stream.

Why is Arm gaining so much share inside data centres?

Power efficiency has become increasingly important as artificial intelligence infrastructure consumes enormous amounts of electricity. Traditional x86 server processors from Intel Corporation and Advanced Micro Devices remain widely used, but cloud companies have increasingly designed custom Arm-based CPUs that can provide strong performance while lowering energy and cooling requirements.

Arm says its architecture represented more than half of server deployments at major hyperscalers during 2025, up from approximately 15% in 2024. Based on current deployments involving Amazon Web Services Graviton, Google Axion, Microsoft Cobalt and Nvidia platforms, Arm expects its share to exceed 60% by the end of 2026.

The AI boom strengthens this opportunity because GPU-heavy servers still require CPUs to coordinate workloads, manage data and run software around accelerator operations. As artificial intelligence moves toward persistent agents that perform multiple software actions between model calls, CPU workload can increase rather than disappear.

Google has begun using Arm-based Axion processors alongside its tensor-processing-unit infrastructure, while Oracle Cloud Infrastructure plans to support Arm’s own AGI processor. These deployments demonstrate that Arm’s data-centre opportunity extends well beyond smartphones, historically the category most closely associated with its architecture.

What exactly is Arm’s AGI CPU and why is it strategically important?

Arm historically designed processor intellectual property and licensed it to semiconductor companies or cloud providers that created the final chips. The AGI CPU changes that relationship because Arm is delivering a complete Arm-designed processor directly to customers rather than stopping at architecture and subsystem IP.

Management launched the product in March with approximately $1 billion of orders and now says confidence has increased in generating as much as $2 billion of AGI CPU revenue across fiscal 2027 and fiscal 2028. Purchase orders have been secured from multiple customers through calendar 2027, while first commercial deliveries remain scheduled for Arm’s fiscal fourth quarter of 2027.

This approach allows Arm to capture a much larger amount of revenue per deployment than it would through royalties alone. A customer buying an entire CPU pays far more than the royalty Arm receives when another manufacturer sells a chip based on Arm architecture.

The trade-off is that Arm takes on responsibilities it historically avoided. Supply commitments, memory procurement, packaging, inventory and customer delivery become far more important when the company sells physical processors rather than only intellectual property.

Does the AGI CPU threaten Arm’s extraordinary gross margins?

Arm reported a GAAP gross margin of 97.1% during fiscal Q2, illustrating the unusual profitability of an intellectual-property-heavy semiconductor model. A complete CPU business is unlikely to sustain the same gross-margin profile because physical products carry manufacturing, memory, packaging and logistics costs.

The initial AGI business remains small relative to Arm’s broader revenue base, meaning the near-term effect on group margins may be limited. As silicon revenue becomes larger, however, investors will need to separate the high-margin IP operation from a more conventional semiconductor product business.

This does not automatically make the strategy unattractive. A lower-margin business can create more absolute profit if it substantially enlarges revenue, particularly where Arm’s architecture provides a differentiated product that customers actively want.

The crucial measure will be return on invested capital rather than gross margin alone. Arm needs AGI CPUs to generate enough incremental operating profit to compensate shareholders for the additional supply-chain and execution risk that comes with moving downstream.

Why did free cash flow drop 73% despite record revenue?

Arm generated only $110 million of non-GAAP free cash flow during the quarter, down from $411 million a year earlier, even though adjusted operating income rose 37%. Operating cash flow declined 63% to $207 million, illustrating how earnings and cash flow can diverge when working-capital requirements and investment timing change.

Research and development spending increased substantially as Arm expanded both its IP roadmap and silicon ambitions. GAAP research and development expense reached $857 million, up 24%, while total GAAP operating expenses increased 27% to $1.2 billion.

The company still generated approximately $1.1 billion of trailing 12-month free cash flow, so one weak quarter does not establish a structural deterioration. The current cash-flow pattern nevertheless provides an early indication of how Arm’s financial model may change as it invests more heavily in complete products.

Investors accustomed to valuing Arm as an asset-light licensing platform need to pay greater attention to operating cash conversion. If AGI CPUs expand rapidly, inventory and supplier commitments could become regular components of the investment case.

Why does the memory shortage matter to Arm now?

A global memory shortage driven by artificial intelligence infrastructure has raised prices and constrained availability across the semiconductor industry. Haas said Arm successfully secured sufficient high-speed DRAM to support its expected $2 billion of AGI CPU revenue, reducing one immediate supply concern.

The fact that Arm now needs to secure memory at all demonstrates how different the new silicon business is from traditional licensing. Under the historical model, Arm customers such as Qualcomm, Apple or hyperscalers managed semiconductor manufacturing and memory procurement themselves.

Direct processors bring Arm into competition for the same components being consumed by Nvidia systems, AMD accelerators and other data-centre hardware. The company must therefore forecast demand accurately enough to secure supply without accumulating excessive inventory.

That problem can become especially difficult when product cycles move rapidly. Buying too little memory constrains revenue, while buying too much can leave Arm carrying inventory if customer orders shift or newer technology arrives sooner than expected.

How serious is the unresolved Qualcomm licensing dispute?

A Delaware jury recently found that Arm breached parts of its licensing agreement with Qualcomm and interfered with Qualcomm’s relationship with Meta Platforms, although damages were only $2. The small damages figure does not mean the wider dispute has disappeared because questions around royalties and future licensing terms remain for judicial resolution.

Qualcomm is an important Arm customer and a major supplier of smartphone and PC processors. Any material change in licensing terms could influence both companies’ economics and set precedents relevant to Arm’s broader customer relationships.

The legal disagreement also highlights a structural tension in Arm’s strategy. As Arm moves toward selling more complete processors and compute subsystems, customers may increasingly view it not only as a neutral technology licensor but also as a potential competitor.

Management needs to expand its economic participation without damaging the ecosystem that made Arm architecture ubiquitous. That balancing act may become one of the most important long-term strategic risks around the silicon transition.

What should investors watch after Arm’s record second quarter?

Royalty growth is the clearest measure because current licence wins need to convert into production shipments. Data-centre royalties doubling is encouraging, but the overall royalty line growing only 8% means the broader conversion still has room to accelerate.

AGI CPU deliveries provide the second test. First commercial shipments during fiscal Q4 2027 should show whether customer orders translate into recognised revenue on schedule and whether product economics remain attractive.

Gross margin and free cash flow will become increasingly important as silicon revenue expands. Investors need to determine whether Arm can move downstream without sacrificing too much of the extraordinary profitability that underpins its valuation.

Arm has already answered the question of whether artificial intelligence creates demand for its architecture. The harder question is whether becoming a chip supplier itself creates more shareholder value than simply collecting increasingly valuable royalties from everyone else building those chips.