On Monday, Intel surprised the world when the company announced that they will manufacture FPGA chips manufactured by relatively unknown player Achronix Semiconductor using 22nm process.
We heard with several sources close to the companies at hand and we were told that one of major reasons why Intel is going to give up to 1% of their 22nm capacity to Achronix is quite simple: Stellarton.
In first quarter of next year, Intel is going to introduce a "fully configurable Intel Atom Processor" codenamed Stellarton. In essence, Stellarton is a dual die package consisting out of Atom E600 processor [45nm Bonnell Architecture] and Altera’s FPGA module.
Next-generation Atom architecture carries a codename Saltwell, while Intel hasn’t revealed the name of the 22nm Atom at this point in time. In order for soon-to-be-introduced Stellarton to have a stellar future in low-power high-density servers, the company needs to get as much experience in the field of 22nm FPGA as possible, and here is where a smaller player such as Achronix comes along.
Intel will manufacture Acrhonix Speedster22i FPGA using the most advanced process when the time is due for introduction of 22nm successor of 45nm Stellarton. Our sources emphasized that Intel does not want to immediately grant 22nm access to FPGA heavy-weights such as Xillinx or Altera, but rather go with a smaller player which might even end up acquired using pocket change. Intel will learn a lot on FPGA with 22nm and Achronix, and instantly reach out to other vendors.
You might ask – ae GlobalFoundries and TSMC business models in danger? Somehow we doubt that. However, if Intel’s stock is to ever reflect on true value of the company, development risks must be shared with other players.
In case you wondered, Achronix is not actively backed by Intel Capital; Argonaut Private Equity, New Science Ventures, Battery Ventures, Entrepia Ventures and Easton Capital make up for the bulk of investment.