Selling a programmable chip means selling against the salesperson's own incentive, and for ten years inside Intel that salesperson was paid to move a Xeon processor first. Altera spent that decade as a business unit told which markets to serve, which foundry to use, and which product to lead with in the bag. The field-programmable gate array, silicon a customer rewires after it ships, kept losing shelf space to the central processing unit sitting next to it.

That is the version of the Altera story worth a CIO's attention, and it is not the version in the wire. Chief Executive Raghib Hussain told Reuters the company is growing roughly 20 percent a year, more than doubling operating income, and preparing for an eventual public listing (Reuters, 2026). The growth number is the announcement. The reason Altera can grow at all is the story.

Independence is a distribution decision before it is a technology one

Silver Lake bought 51 percent of Altera for $4.46 billion last September, a transaction valuing the company at $8.75 billion, with Intel keeping a passive 49 percent (Reuters, 2026). Set that against the $16.7 billion Intel paid in 2015 (Intel, 2015). Intel wrote down roughly half the value of an asset it once called central to its data center ambition.

Hussain has been direct about why the asset stalled. Under Intel, Altera had to focus on telco and data center markets, use Intel's foundry, and rely on a sales team more motivated to sell processors than programmable chips (EE Times, 2025). None of those constraints was a comment on FPGA demand. They were org-chart decisions that starved a product line inside a company fighting for its own survival.

Independence removes the constraints in one move. Altera now picks its own foundry, sets its own roadmap toward embedded, aerospace and defense, audio and video, robotics, drones, and edge artificial intelligence, and fields a sales team paid to sell one thing.

The pitch is clean. The execution record is not.

The nervous-system bet asks buyers to fund a decade the category has not earned

Hussain's framing for where FPGAs go next is memorable. "If GPU is the brain, the FPGAs are the nervous system," he told Reuters, projecting FPGA content of $100 to several hundred dollars per robot and a market worth 100 billion to several hundred billion dollars over a decade (Reuters, 2026). The role he describes is real: connectivity, data pre-processing, and sensor fusion sitting alongside the graphics processing units from NVIDIA and others that do the heavy inference.

Here is the tension a buyer has to hold. Altera's own revenue fell from $2.9 billion in 2023 to $1.5 billion in 2024 (Reuters, 2026). Across the aisle, the embedded segment at AMD, which is mostly the Xilinx FPGA business it bought in 2022, dropped 33 percent in 2024 to $3.6 billion (Motley Fool, 2025).Both lost significant revenue in the same year the robotics thesis is supposed to be taking off, AMD's embedded segment by 33 percent and Altera by roughly half.

A projection of hundreds of billions over a decade sits on top of a category that just contracted. That gap does not make the thesis wrong. It makes it a bet on a turn that has not shown up in a single year of shipping numbers yet.

The roadmap runs through Taiwan, and that is the part a continuity plan has to price

Altera manufactures with both Intel Foundry and TSMC, and its leading-edge products are moving to TSMC's 2-nanometer and 3-nanometer technology (Reuters, 2026). A newly independent United States chip maker, pitched partly on aerospace and defense demand, ties its most advanced roadmap to fabrication in Taiwan. Any CIO who watched supply concentration become a board-level topic already knows how that sentence reads in a risk review.

Hussain also cut headcount by 82 around independence and inherited a brand that, in his own words, lost trust in its excellence, quality, and schedule over the prior decade (EE Times, 2025). He has reduced Intel transition agreements from 125 to 15 and shipped six prototype chips in a year (Reuters, 2026). Those are signs of a team executing. They are not yet proof of a company that ships at scale, on schedule, for a decade.

The FPGA is a long-cycle design-in. A chip specified into a robot or a defense system in 2026 is a dependency that has to hold through 2030 and beyond.