Intel turns in its strongest report in 15 years: AI compute demand spreads from GPUs to CPUs, and does the semiconductor cycle enter a new stage?

On the U.S. stock market’s after-hours session on July 23, 2026, Beijing time, Intel (INTC) released its 2026 fiscal second-quarter earnings report for the period ended June 27. The significance of this earnings release was clear the moment the data was published—revenue of $16.1 billion, up 25% year over year, the fastest single-quarter revenue growth rate since Q3 2011. Adjusted EPS of $0.42 doubled the market expectation of $0.21. Non-GAAP gross margin was 41.8%, up 12.1 percentage points significantly from the prior-year period.

After the earnings were released, Intel’s stock price surged more than 13% at one point in after-hours trading. But in the following regular trading session (July 24, Beijing time), Intel closed at $100.23, down 2.33% on the day—between that upswing and that downswing, the market’s pricing of this “above-expectations” earnings report was not one-directionally optimistic; differences in views remain.

What’s truly worth asking is: what exactly does this Intel earnings report mean? Over the past two years, the spotlight of the AI boom has almost entirely been on Nvidia, AMD, and Broadcom. GPUs have been viewed as the only narrative for AI compute. But Intel’s data center business—$6.3 billion in revenue, up 59% year over year—is rewriting this story. Demand for AI infrastructure is spreading from GPUs to a broader chip supply chain.

Data center: From “supporting role” to growth engine

In the second quarter, Intel’s Data Center and AI (DCAI) segment generated revenue of $6.3 billion, up sharply 59% year over year, and up 24% quarter over quarter, marking the strongest server business quarterly growth in the company’s history. This growth rate was more than twice Intel’s overall revenue growth rate, and comfortably beat analysts’ expectation of $5.6 billion.

In the earnings call, Intel CEO Pat Gelsinger said plainly: “In the data center, the CPU is taking off.” He added that demand is outpacing Intel’s continuously increasing supply capacity. Intel CFO David Zinsner further noted that from a pricing perspective, the situation is better than the company expected—in the China market, the cumulative price increase for some server CPU products since the beginning of 2026 has already exceeded 40%.

The structural forces driving this growth lie in the shift of focus within the AI industry. Over the past two years, the core market attention has been on high-performance GPU-accelerated chips needed to train AI models. But as enterprises move from training large models to large-scale deploying “agentic AI” that can execute tasks autonomously, compute demand on the inference side begins to explode. At this stage, AI servers no longer only need GPUs—CPUs, as the core hub for system orchestration, see demand surge as well.

TrendForce research indicates that AI applications are shifting from training toward inference and agentic AI, upgrading server CPUs from auxiliary components to the core orchestration hub. Intel’s Xeon 6 series processors have become one of the fastest ramping products in the company’s history. Meanwhile, AMD’s EPYC server processors also benefit from this trend, with analysts expecting AMD’s 2026 server CPU shipments to grow 15% to 20%.

In other words, an AI data center is a complex system made up of GPUs, CPUs, network chips, and storage. As compute demand spreads from the training side to the inference side, every link in the supply chain benefits.

Is Intel regaining market confidence?

Beyond the headline earnings figures, the market cares most about the sustainability of Intel’s recovery this time.

From a fundamentals perspective, there are several positive signals. First, Intel has exceeded its financial guidance for the seventh consecutive quarter, and has achieved profitability under Non-GAAP measures for four straight quarters. Second, the company has signed 10 long-term supply agreements with server CPU customers—some lock in pricing, some lock in purchase volumes, and some cover timeframes as long as three to five years. This long-term contract model provides greater predictability for future revenue.

Third—and most importantly—Intel is regaining pricing power. A tight supply situation shows up not only in sales volumes, but also in prices. Zinsner clearly said that “pricing is better than expected,” and some CPU products’ cumulative price gains of more than 40% since the beginning of the year are the most direct evidence for that assessment.

But risks also exist. Intel Foundry’s second-quarter revenue was $5.8 billion, up 31% year over year, but operating loss remained $2.1 billion. More importantly, the foundry’s orders currently mainly come from Intel’s own product divisions rather than external customers. Although cybersecurity company Fortinet became the first publicly disclosed foundry customer since Pat Gelsinger took office, this partnership uses more mature manufacturing processes rather than Intel’s most advanced 14A node.

On the advanced process front, Intel’s 18A node has entered mass production and its yield is above expectations, and the R&D progress for the next-gen 14A process is in line with expectations. The company has said it is “fully committed” to ramping mass production using the 14A process in 2028. But whether the foundry business can attract enough external large customers remains a key variable determining whether Intel’s long-term transformation can succeed.

Semiconductor cycle: From compute shortages to infrastructure expansion

If you zoom out from Intel alone to the entire semiconductor industry, a bigger question emerges: is the semiconductor cycle entering a recovery phase?

From 2024 to 2025, AI compute shortages were the industry’s main storyline—GPU demand outstripped supply, HBM memory prices surged, and advanced packaging capacity faced shortages. Entering 2026, the narrative is changing. Major global cloud providers and AI companies continue to raise budgets for AI infrastructure investment. Nomura estimates that global AI server revenue will grow 78% in 2026 and 76% in 2027. WSTS expects the global semiconductor market size to grow by more than 25% in 2026.

Intel management’s view in the earnings call was more specific: it expects that future server CPU industry shipment volumes will maintain high double-digit growth in 2026 and 2027, with the upcycle extending through 2028. Gelsinger emphasized that the industry’s core contradiction today is “demand far exceeding supply,” with long-term shortages across four areas—wafers, substrates, memory, and advanced packaging—and that these cannot be alleviated in the short term.

Based on this assessment, Intel announced that it would raise its 2026 capital expenditure guidance to more than $20 billion and indicated that 2027 capex would be significantly higher than 2026. The company disclosed that from 2021 to 2026, its U.S. tooling and facilities capex totaled close to $100 billion.

In terms of cycle positioning, the semiconductor industry is currently in a stage where “inventory clearing” and “capacity expansion” coexist. Soaring inference-side demand for AI has become the core variable driving a simultaneous increase in volume and pricing for the chip design segment. The industry is transitioning from “higher unit prices” to “capacity expansion.”

Of course, an upcycle does not mean there are no risks. A recent report from Morgan Stanley warned that the AI-driven semiconductor storage industry celebration is approaching a turning point, and memory contract prices are expected to peak in Q4 2026. Memory chip price trends are often viewed as a leading indicator for the semiconductor cycle—if storage prices peak first, does it imply that the broader chip cycle may also face a turning point? That’s a variable that needs ongoing tracking.

Conclusion

Intel’s 2026 second-quarter earnings report is not only a company-level earnings beat, but also a landmark signal that AI compute demand is spreading from GPUs to a broader chip supply chain. The “takeoff” of data center CPUs, the return of pricing power, and the sharp increase in capital expenditures together point to one judgment: the AI chip cycle is moving from the “compute shortage” stage into a new “infrastructure expansion” phase.

Of course, the evolution of a cycle is never linear. Expanding external customers for the foundry business, a potential turning point in storage prices, and the sustainability of capital expenditures by tech giants are all uncertain factors that require continued monitoring. But for Intel and the broader semiconductor supply chain, AI demand spreading across the full chain—from GPUs to CPUs, network chips, and storage—is opening up a larger growth space than a pure “GPU narrative.”

FAQ

Q1: What is the most core above-expectations data in Intel’s Q2 earnings report?

Intel’s second-quarter revenue was $16.1 billion, up 25% year over year, the fastest growth rate since Q3 2011, far above the market expectation of $14.4 billion. Adjusted EPS of $0.42, double the market’s expected $0.21. Data center segment revenue was $6.3 billion, up 59% year over year. The company has exceeded its financial guidance for seven consecutive quarters.

Q2: Why is Intel’s CPU business growing so fast suddenly?

The AI industry is shifting from training models to large-scale deploying “agentic AI.” Exploding compute demand on the inference side upgrades general-purpose CPUs from auxiliary components to the core orchestration hub for AI servers. AI data centers not only need GPUs, but also large amounts of CPUs to support server, storage, network, and inference workloads. Intel’s Xeon 6 series has become one of the fastest ramping products in the company’s history.

Q3: Can Intel’s recovery this time be sustained?

The company expects server CPU industry shipment volumes to maintain high double-digit growth in 2026 and 2027, and that the favorable cycle will extend to 2028. Intel has raised 2026 capex to more than $20 billion and expects 2027 to continue to increase sharply. But whether the foundry business can attract enough external large customers remains a key variable. The risk that storage prices could peak in Q4 2026 also needs attention.

Q4: What stage is the AI chip cycle in right now?

The industry is moving from the “AI compute shortage” stage of 2024 to 2025 into a new phase of “AI infrastructure expansion.” Major global cloud providers continue to raise their AI infrastructure investment budgets, and Nomura expects global AI server revenue to grow 78% in 2026. The current semiconductor industry is in a phase where inventory clearing and capacity expansion coexist, and a surge in inference-side demand is the core variable driving volume and pricing increases across the chip design segment.