Intel's Q2 2026 results gave investors the comeback number they wanted: $16.1 billion in revenue, its fastest quarterly growth since 2011. The harder question is whether Intel can actually build enough chips to turn that demand into a durable recovery.
For most of the last decade, Intel was the cautionary tale of a technology giant that missed the turns. It missed mobile, stumbled through manufacturing delays, and watched Nvidia become the defining company of the AI era while it tried to make its own foundry plan credible again. So when Intel reported Q2 2026 earnings on July 23, the numbers landed with some force: $16.1 billion in quarterly revenue, up 25% from a year earlier, and adjusted earnings of $0.42 a share, nearly double the Wall Street estimate of about $0.21 to $0.22. Analysts had expected roughly $14.4 billion in revenue. Intel beat that by about $1.7 billion. That is not a routine beat.
According to Intel's own results release, this was the company's strongest revenue growth in more than fifteen years. Its Data Center and AI segment did the heavy lifting, with revenue rising 59% year over year to $6.3 billion. CEO Lip-Bu Tan said AI is driving unprecedented demand for compute, and Intel said the quarter was helped by stronger execution across CPUs, ASICs, advanced packaging and its foundry network. You can dismiss plenty of chip-company AI language as market theatre. This one is harder to wave away, because the revenue moved with it.
Intel still posted a GAAP loss attributable to the company of $11 billion, or $2.16 a share. Don't confuse this with a clean victory lap. The formal accounts still show the cost of the rebuild - the non-GAAP profit is what investors traded on after the print, but the damage from years of expensive fixes is right there in the numbers. That is the real Intel story right now.
Lip-Bu Tan became Intel's CEO in March 2025, not early 2024, after the board replaced Pat Gelsinger with interim leadership and then handed Tan the job. His first 16 months have been about focus and cuts. Business Insider recently noted that Intel has also been cutting in its data center group, even as that unit is now producing the company's sharpest growth. That looks odd at first. It isn't. It tells you Tan is still trying to shrink the old Intel while feeding the parts of the company that customers actually need.
Intel shares jumped after the results, with reports from the Wall Street Journal and Barron's putting the after-hours move at roughly 12% to 13%. The stock had already risen sharply this year before earnings, so the reaction wasn't simply relief. Investors were paying for evidence that Intel has something scarce again.
The shortage is the signal #
Here's the thing: Intel says demand is running ahead of supply. Investing.com reported that Intel disclosed 10 long-term supply agreements with data center customers, while management said the company remains supply constrained. That is a better problem than empty fabs, but it is still a problem. Customers wanting more chips than Intel can produce doesn't help you much if your factories can't catch up.
Still, a 59% data center gain while supply is tight changes the frame. It means the current number may not be the ceiling. It may be the first visible piece of pent-up demand. Intel also raised its 2026 capital expenditure forecast to more than $20 billion, up from $18 billion, a plain sign that management expects the bottleneck to matter for more than one quarter.
The foundry side gives you the same mixed picture. Intel Foundry revenue rose 31% to $5.8 billion, according to the company release, and MarketWatch reported that Counterpoint Research's Neil Shah pointed to Intel's 18A process reaching an 85% yield. That is a useful datapoint because 18A is central to Intel's attempt to claw back manufacturing credibility. But useful is not the same as finished. Intel still has to turn better yields, advanced packaging demand and named customer wins into profitable outside foundry revenue.
Fortinet is one of those wins. Investors.com reported that Intel's momentum included Fortinet as a foundry customer, and Tom's Hardware said the cybersecurity company is working with Intel on its next-generation Security Processor 6 using Intel 4. That is not the kind of massive advanced-node hyperscaler win that would suddenly put Intel beside TSMC. It is still real. You need real customer names before you get the bigger claims.
Intel is selling a different AI story #
Nvidia still owns the market's imagination around AI training. That hasn't changed. But Intel's stronger pitch is not that it has suddenly become Nvidia. It is that AI infrastructure needs more than GPUs, especially as inference workloads put pressure on CPUs, packaging, memory and full rack systems. MarketWatch, citing FactSet estimates, noted that Intel's Data Center and AI revenue of $6.3 billion came in well above the expected $5.4 billion. That is where you should look, not at a fantasy head-to-head with Jensen Huang's company.
For Q3, Intel guided revenue of $15.8 billion to $16.8 billion and non-GAAP EPS of $0.38, ahead of Wall Street expectations cited by multiple outlets. The midpoint implies continued double-digit growth. It also leaves management room in case supply remains tight or memory costs bite harder later in the year. Fifteen years is a long time to wait for your fastest quarter. Intel has earned the right to be taken seriously again, but not the right to be believed without proof. The next proof is physical: more wafers, more shipped server CPUs, more outside foundry customers, and fewer excuses about capacity.
Also read: A $700 Billion Sovereign Fund Just Made the Chinese AI Cost Argument Impossible to Ignore • Meta and BlackRock's $12 billion El Paso bond deal shows the AI arms race is finally moving the credit market • China is winning the Global South's AI future by giving models away for free