Innovation & Tech

Intel Broke a 26-Year Curse

Intel stock is up 445% in twelve months. A year ago, analysts were writing post-mortems. The company had lost

By Conor Healy

June 8, 2026

Intel stock is up 445% in twelve months. A year ago, analysts were writing post-mortems.

The company had lost 60% of its value in 2024. Its CEO had been ousted. Its foundry business was bleeding cash. Nvidia and AMD had lapped it in the AI race. The stock traded below $18.

Then it hit $133, an all-time high. Intel had not traded at that level since August 2000, during the dot-com bubble. The 26-year curse was broken.

What happened in between is one of the stranger turnaround stories in recent tech history.

Lip-Bu Tan

Pat Gelsinger was removed as CEO in December 2024 after Intel’s stock collapsed. His replacement was Lip-Bu Tan, a veteran investor and semiconductor executive who had previously run Cadence Design Systems.

Tan inherited a company with structural problems. Intel had fallen behind TSMC in manufacturing technology. Its foundry division, meant to compete with TSMC for external customers, had no major wins and was losing billions annually. Its AI chips were an afterthought compared to Nvidia’s GPUs.

Within months, Tan restructured the organisation. Management layers were compressed from twelve to five. Engineering leaders began reporting directly to the CEO. A “bad news first” policy required problems to surface within 24 hours. Unnecessary meetings were eliminated. According to Bloomberg, some decisions that previously took a year now took days.

The changes were cultural as much as structural. Intel had become slow, and Tan’s job was to make it faster.

18A

The technical centrepiece of Intel’s turnaround is a manufacturing process called 18A.

Semiconductor manufacturing is measured in nanometres, though the numbers no longer correspond directly to physical dimensions. What matters is density and efficiency. Intel’s 18A process is competitive with TSMC’s most advanced nodes, which means Intel can theoretically manufacture chips as good as anything TSMC produces.

This matters because TSMC manufactures chips for Apple, Nvidia, AMD, and most of the industry. If Intel’s 18A works at scale, Intel becomes an alternative. American companies gain a domestic option for advanced manufacturing. The geopolitical implications are significant given Taiwan’s proximity to China.

Intel announced approximately 200 design wins tied to 18A. The company’s roadmap extends to 14A, 10A, and 7A, with 14A risk production targeted for 2028. Whether Intel can execute remains uncertain, but the trajectory is no longer theoretical.

$5 Billion from Nvidia

Three external investments signalled that Intel’s turnaround was being taken seriously.

Nvidia invested $5 billion in September 2025. The optics were striking: the company that had dominated AI chips was putting money into its struggling competitor. The investment was financial rather than operational, and Nvidia has reportedly paused its trial of Intel’s 18A process. But the capital injection provided credibility at a critical moment.

The Trump administration directed CHIPS Act funding toward Intel’s domestic manufacturing expansion. Government support de-risked the foundry buildout and aligned Intel with national security priorities around semiconductor independence.

Elon Musk’s “Terafab” project, a massive AI infrastructure initiative, included Intel as a partner. The details remain vague, but the association connected Intel to the most ambitious compute buildout currently underway.

None of these guaranteed success. All of them suggested that serious money believed success was possible.

Training vs Inference

The bull case for Intel rests on an argument about where AI workloads will run.

For the past several years, AI has been synonymous with GPUs. Nvidia’s chips train large language models. The entire AI boom has been built on GPU infrastructure. Intel, a CPU company, was largely irrelevant to this.

But training is only part of the AI lifecycle. Inference, the process of running trained models to generate outputs, can happen on CPUs. As AI shifts from building models to deploying them, the workload mix may shift too.

Intel’s Xeon server CPUs are already seeing surging demand in AI data centres. If the ratio of inference to training increases, and if inference favours CPUs over GPUs, Intel’s core product line becomes relevant to AI in a way it has not been.

This is not guaranteed. Nvidia is building inference chips. AMD is competitive on both CPUs and GPUs. But the argument gives Intel a path into the AI market that does not require beating Nvidia at its own game.

Priced for Perfection

Intel stock is now priced for near-flawless execution.

The foundry business is still losing money. Gross margins remain below target. The 18A process has design wins but limited production data. Apple is reportedly in preliminary discussions to manufacture chips at Intel, but no deal has been confirmed. Nvidia’s decision to pause its 18A trial suggests the technology is not yet where it needs to be.

Competition remains intense. AMD has taken CPU market share for years and continues to press. Arm-based chips are gaining traction in data centres. TSMC’s manufacturing lead, while narrowing, has not disappeared.

A stock that has risen 445% in twelve months has priced in a lot of good news. If execution stumbles, the downside is significant.

Rally or Revival

Intel’s all-time high is not just a stock story. It reflects a broader shift in how the market views AI infrastructure.

For years, the assumption was that AI would be built entirely on GPUs, that Nvidia would capture most of the value, and that legacy chipmakers like Intel were irrelevant. The assumption is now being revised.

CPUs may matter more than expected. Domestic manufacturing may be strategically valuable. A company left for dead can, under the right leadership, become competitive again.

Intel has not proven it can sustain the turnaround. The foundry business is still unprofitable. The 18A process has design wins but limited production data. The Apple deal remains unconfirmed. The CPU-to-GPU ratio in AI workloads may or may not shift in Intel’s favour.

But for a company trading below $18 a year ago, reaching an all-time high is already an outcome few predicted. The next twelve months will determine whether it was a rally or a revival.