Company Semiconductors · Santa Clara, CA
The company that built the microprocessor is trying to build a second act
The company that built the microprocessor is trying to build something harder: a second act. Inside Intel's bet-the-house pivot from selling chips to making everyone else's.
In the autumn of 1971, Intel put an advertisement in a trade magazine announcing "a new era of integrated electronics." The product was the 4004, a sliver of silicon the size of a fingernail holding 2,400 transistors. It could do arithmetic faster than a machine that once filled a room. Fifty-five years later, the company that shrank the computer onto a chip is trying to pull off something that has proven far more difficult than any transistor: reinventing itself while the whole industry watches.
Intel today is a $52.85 billion business with roughly 85,000 employees and headquarters at 2200 Mission College Boulevard in Santa Clara, California - the heart of the valley that silicon named. It still designs the processors inside a large share of the world's laptops and servers. But the story that matters now is not the chips it sells. It is the factories it owns, and the decision to open them to the rest of the industry.
What Intel actually does
Strip away the jargon and Intel does two things most of its rivals gave up trying to do together. It designs chips, and it manufactures them. Companies like AMD, Apple and Nvidia are "fabless" - they draw the blueprints and hand them to a factory in Taiwan or South Korea. Intel is an integrated device manufacturer: it runs its own fabrication plants, or "fabs," in Arizona, Oregon, New Mexico and Ohio.
The product catalogue splits cleanly. Intel Core and the newer Core Ultra processors sit inside consumer PCs, with the Core Ultra line adding a dedicated neural processing unit for on-device AI - the pitch behind the "AI PC." Xeon processors run data centers and the cloud. Arc is Intel's line of discrete graphics cards for gamers and creators. Gaudi accelerators chase the generative-AI training market. And FPGAs, from the Altera side of the house, handle the reprogrammable logic that networking gear and embedded systems rely on.
Who buys it
The customers fall into two very different rooms. In one are the PC makers - Dell, HP, Lenovo - plus the cloud operators, enterprises, gamers and the makers of embedded and IoT devices who have bought Intel silicon for decades. Billions of devices carry it. In the other, newer room sit the customers who used to be competitors: other chip companies who might pay Intel to build the chips they designed. That room was empty for years. In 2026 it got its first named occupant.
Only the paranoid survive. Andy Grove, former Intel CEO
The problem it is trying to solve
For most of its history Intel's problem was other people's: how to make computers faster and cheaper. The problem it is solving now is its own. Through the late 2010s and early 2020s, Intel lost its manufacturing lead to Taiwan's TSMC, watched Arm-based designs eat into its markets, and stumbled on its own process roadmap. In 2024 the company parted ways with its CEO. It has since cut its workforce by roughly a quarter - from about 108,900 people to 85,100 - and posted a net loss for FY2025.
The strategic answer is Intel Foundry. If the world worries about relying on a handful of factories in one geography, Intel argues, then a leading-edge fab network on American and European soil is worth paying for. The United States agreed: through the CHIPS Act, the Commerce Department committed up to roughly $7.86 billion in direct funding to expand Intel's domestic manufacturing.
How it is different
The difference is the thing that looked like a liability. Keeping the fabs cost Intel dearly while fabless rivals stayed light. But manufacturing at the leading edge is one of the hardest and most capital-intensive things humans do, and only a tiny number of companies on Earth can do it at all. Intel is the only one headquartered in the United States. In a decade defined by supply-chain anxiety and export controls, "we make it here" became a feature.
The technical expression of that bet is 18A - a manufacturing process built on two innovations Intel likes to name-check: RibbonFET, a gate-all-around transistor, and PowerVia, which routes power to the chip from the back. It is the node the whole turnaround rides on. If 18A is competitive and yields well, the foundry pitch is real. If it slips, the story gets harder.
The business model, plainly
Intel makes money three ways, which map to its three segments. The Client Computing Group sells the CPUs in your laptop. Data Center & AI sells the Xeons, GPUs and networking chips that run the cloud. And Intel Foundry - the newcomer - sells manufacturing capacity to outside customers, turning what used to be a pure cost center into a business with its own income statement. It is a deliberate copy of the model that made TSMC the most valuable company in the industry, attempted by the one Western firm with the fabs to try.
This is going to be a marathon, not a sprint. Lip-Bu Tan, Intel CEO
The man holding the hand
In March 2025 Intel named Lip-Bu Tan chief executive. A veteran of the semiconductor industry with a long investing and operating record, he inherited a legend in decline and a factory network worth tens of billions. His playbook has been to simplify: cut headcount, focus the roadmap, and sell the foundry to anyone who will listen. Investors have, so far, listened back - Intel shares rose more than 300% in the period following his appointment, a vote less on the present numbers than on the pivot.
By July 2026 that pitch had a proof point. Fortinet agreed to build its next-generation SP6 security chip at Intel Foundry on the Intel 4 process - the first publicly named customer. Reports placed Tesla, Google, Microsoft and even Apple among interested or committed parties. Intel also signaled it was pushing deeper into GPUs, hiring a new chief architect to lead the effort. None of it guarantees the comeback. All of it is the shape a comeback would take.
The expertise, and the culture
What Intel has that money cannot quickly buy is process knowledge - the accumulated craft of turning sand into working logic at the scale of a few nanometers, where a single stray atom can ruin a transistor. That expertise runs from lithography and materials science through packaging, where Intel has invested heavily in stacking chips vertically to squeeze out more performance. It also runs through software: compilers, drivers, the OpenVINO and OneAPI toolchains, and decades of x86 tuning that developers quietly depend on.
The culture that built all of it was set early by Noyce, Moore and Andy Grove. Grove's "constructive confrontation" - argue hard, decide, commit - and his warning that "only the paranoid survive" became management folklore well beyond Silicon Valley. That same intensity is now aimed inward, at a restructuring large enough to reshape the company. Whether Intel can keep its engineering depth while shedding a quarter of its people is one of the open questions of the turnaround.
Where it fits in the market
Intel sits at an unusual crossroads: it competes with everyone and, increasingly, wants to supply them too. In CPUs it faces AMD and the Arm ecosystem - Qualcomm, Apple and others. In GPUs and AI accelerators it trails Nvidia and AMD. And in manufacturing, the business it is staking its future on, it is chasing TSMC and Samsung. Being the challenger in every race at once is a strange place for the company that once set the pace in most of them. It is also, arguably, the most interesting corner of the chip industry to watch right now.
What you can do with it
For most people, Intel is invisible infrastructure - the reason a laptop wakes instantly or a cloud service answers in milliseconds. For developers and enterprises it is more hands-on: open-source toolkits, the OpenVINO framework for running AI models efficiently, and the OneAPI stack for programming across CPUs, GPUs and accelerators. For a chip startup, the new option is the most consequential - the ability to have a leading-edge design built in the United States without owning a fab. That, more than any single processor, is what Intel is selling now.