Intel Reconsiders FPGAs

Data-Center Synergy Fails to Materialize; Other Markets Underserved

Bryon Moyer

Eight years after acquiring FPGA maker Altera, Intel is having a change of heart. It’s spinning its Programmable Systems Group into a separate unit with a loose plan to publicly list the new company in a couple of years while keeping a controlling stake.

As of January 1, 2024, Intel will start reporting separate results for the as-yet-unnamed new division (Altera?). Sandra Rivera, currently executive VP of the Data Center and AI group, will head the new entity, which will set its own strategic direction, design its own products, and assemble its own salesforce untethered from Intel’s strategy and sales channels. It will continue using Intel as a foundry.

In 2015, when Intel acquired Altera for $16.7 billion, it viewed the company as a key enabler for moving FPGAs—and Intel as a whole—further into the data center, combining Xeon CPUs and FPGAs in a single package. The first such device (the Xeon Gold 6138P) failed to catch on, however, and was discontinued.

Meanwhile, the company has missed out on FPGA applications outside the data center, leaving them primarily to AMD (which acquired Altera archrival Xilinx in 2022), Lattice, and Microsemi. The move allows the new company to address those neglected areas without defocusing Intel. It also raises the question as to whether the change reflects unique Intel challenges or whether AMD may come to the same conclusions regarding its FPGA offering.