Intel reported first-quarter 2026 earnings on April 23 that blew past Wall Street’s expectations by a margin that made analysts do a double take.
It sent the stock surging approximately 20 percent in after-hours trading and pushing it to an all-time high, clearing a price ceiling that had held since the dot-com era more than two decades ago.
The headline numbers were remarkable.
The reason Intel stock is doing what it is doing right now is a story about what artificial intelligence is actually doing to the chip industry, and it is not the story most people have been telling about AI and semiconductors for the past three years.
The Numbers Behind Intel’s Success
Intel reported Q1 2026 revenue of $13.58 billion against a Wall Street consensus of $12.42 billion, a beat of more than $1.16 billion, or roughly 9.4 percent above expectations. Revenue grew 7.2 percent year-over-year.
Non-GAAP earnings per share came in at $0.29 against an analyst consensus of approximately $0.01 to $0.02, a beat so large that one publication described it as a 1,350 percent earnings surprise.
Non-GAAP net income was $1.5 billion, more than double the $0.13 per share Intel earned in the same quarter a year ago.
It was Intel’s sixth consecutive quarter of revenue above its own guidance. The company’s own January outlook for Q1 had been $11.7 billion to $12.7 billion. The actual result came in $1.4 billion above the top of that range.
The Q2 guidance extended the surprise. Intel guided Q2 revenue at $13.8 billion to $14.8 billion against a consensus of approximately $13.1 billion, another massive beat at the guidance level. Q2 non-GAAP EPS guidance was $0.20 against a consensus of about $0.09 to $0.10.
There is a GAAP loss in the Q1 results, a net loss of $3.7 billion, or $0.73 per share. That number is not representative of Intel’s operating performance.
It is almost entirely driven by non-cash charges: $4.1 billion in restructuring items including a Mobileye goodwill impairment, plus $1.09 billion in mark-to-market losses on Escrowed Shares.
The non-GAAP numbers, which strip out these non-cash items, are what reflects the actual business trajectory, and that trajectory is sharply upward.
Intel’s Three Main Segments
Intel’s business breaks into three main segments. Client Computing, PC chips, brought in $7.7 billion, up 1 percent year-over-year and above internal expectations.
The Data Center and AI segment brought in $5.1 billion, up 22 percent year-over-year and 7 percent sequentially, with operating profit of $1.5 billion, a 31 percent margin.
Intel Foundry brought in $5.4 billion, up 16 percent year-over-year, though it is still running at an operating loss that narrowed to $2.4 billion from a trough of $4 billion-plus at its worst point in 2025.
The number that tells the structural story is buried in the Client Computing segment. AI PC revenue grew 8 percent sequentially and now represents more than 60 percent of Intel’s client CPU mix.
More than six in ten PCs using Intel processors are now AI PCs. ASIC revenue, custom chips, nearly doubled year-over-year.
Why AI Is Helping Intel
For the past three years, the dominant semiconductor narrative has been: AI is good for Nvidia, bad for everyone else. The logic was straightforward.
Training large AI models requires massive GPU clusters. Nvidia makes the GPUs. Intel makes CPUs. CPUs were not the training chip. Intel lost.
CEO Lip-Bu Tan is arguing, and the Q1 numbers are beginning to support, that this framing missed the second half of the story.
“The CPU is reinserting itself as the indispensable foundation of the AI era,” Tan said on the earnings call. “This isn’t just our wishful thinking, it’s what we hear from our customers.”
The argument works like this. AI is moving from training, where you build the model once using massive GPU clusters, to inference and agentic workloads, where you actually use the model continuously to do things.
Inference is different from training. It is latency-sensitive. It is distributed. It happens at the edge, in data centers, inside PCs, across server racks that already have CPUs in them.
Every data center in the world runs on Intel x86 architecture. Every PC does. Every edge server does.
Tan said on the call that customers are now deploying server CPUs alongside GPU accelerators at ratios that are “moving back towards CPU,” meaning the GPU-dominant architecture of the training era is giving way to a more balanced CPU-GPU ratio in the inference and agentic era.
Intel’s data center CPU revenue being up 22 percent year-over-year is the number that reflects this shift in real dollars.
Intel signed a long-term agreement with Google in Q1 for volume and pricing of server CPUs and ASICs, a contract typically lasting three to five years. Tesla and SpaceX are confirmed customers for Intel’s 14A process node.
Tan specifically named Elon Musk, “As we look to continue challenging the status quo, I can think of no better partner than Elon Musk.”
A new collaboration with SambaNova on AI inference architecture was also announced alongside the results.
The Dot-Com Ceiling
Intel’s stock surging to an all-time high, clearing a price level not seen since the dot-com era, is a remarkable thing to write. The company spent more than two decades below its late 1990s highs.
It cut 15 percent of its workforce in July 2025. It cancelled fabrication plants in Germany and Poland.
Its foundry business was hemorrhaging money. Its stock had significantly underperformed every major chip competitor for years.
The recovery began in 2025, the stock soared 84 percent that year. Then it was up more than 80 percent further in 2026 year-to-date, heading into earnings.
By the time Thursday’s closing bell rang, Intel had more than tripled over the trailing twelve months.
That run-up had created a forward price-to-earnings multiple of nearly 100 times expected earnings, a valuation that, as Sherwood News noted, was actually exceeding the lofty multiples Intel received during the peak of the dot-com boom itself.
The market had priced in an enormous amount of recovery. The Q1 numbers suggest it was not wrong to do so, because the actual earnings came in so far above what anyone expected that the multiple now looks considerably less stretched.
When you price a stock at 100 times $0.02 in expected earnings, it looks insane. When the company actually earns $0.29, the math changes.
Intel’s stock has cleared its dot-com-era ceiling. The question the market is now wrestling with is whether this is the beginning of a genuine structural revival, or an exceptional quarter in the middle of a trend that has further to run, or has already run too far.
Tan’s answer, delivered six consecutive times now via quarters that beat his own guidance, is that the CPU is back. The AI era needs inference. Inference needs CPUs. Intel makes CPUs. That is the trade.
