Intel and Altera extend foundry deal into interposer and full 3D

Altera has extended its foundry agreement with Intel to cover future 3D-IC development, effectively quashing rumors that it was about to move its most advanced production back to TSMC.

Intel originally had a deal to produce Altera’s 14nm Stratix 10 FPGAs based on its TriGate finFET process. The new agreement covers “collaboration on the development of multi-die devices”.

“Altera’s work with Intel will enable the development of multi-die devices that efficiently integrates monolithic 14nm Stratix 10 FPGAs and SoCs with other advanced components, which may include DRAM, SRAM, ASICs, processors and analog components, in a single package. The integration will be enabled through the use of high-performance heterogeneous multi-die interconnect technology,” a statement says.

It then adds that, “Intel and Altera are currently developing test vehicles aimed at streamlining manufacturing and integration flows.”

What is potentially significant about that is that it mirrors work Altera has been undertaking with TSMC already, and for some time. In March 2012, the two companies unveiled the world’s first heterogeneous test structure based on TSMC’s CoWoS (chip-on-wafer-on-substrate) interposer technology.

So, is TSMC completely out of the picture with regards to Altera’s cutting edge devices – the foundry giant still produces pre-finFET devices for Altera? Quite possibly not.

When Altera originally signed with Intel in February 2013, TSMC was quick to point out that it retained a strong supply relationship with the FPGA company. We won’t be surprised if something similar follows in the next few days, perhaps even hours.

More to the point, the Intel-Altera statement makes no mention of this being an ‘exclusive’ deal for 3D. Given the challenges that technology has presented, and the existing knowledge bank on 3D that Altera has built up with TSMC, it may make sense for Altera to dual source the technology.

Moreover, TSMC has been making the running in the industry in terms of showcasing various interposer options that it is bringing to market (although, yes, when has Intel ever done that much publicly).

The rumor that Altera and Intel were to split, originating in the Asian press, was largely based around delays to Intel’s introduction of 14nm, and the prospect that it would let Xilinx beat Altera to the punch with the next generation of FPGAs.

In that light, the big winner for the moment is Intel. Altera’s vote of confidence will help remove some of the question marks over Intel’s foundry operation – still, despite this deal, in need of more customers – by helping to seed broader confidence.