Intel announces $1bn fund for Intel Foundry Services
16-02-2022 | By Robin Mitchell
Recently, Intel announced that it has prepared a $1bn investment fund aimed at helping companies to use Intel services for manufacturing next-generation CPU technologies, including x86, ARM, and RISC-V. What exactly does the announcement from Intel outline, why is this a major shift in business tact for Intel, and what could it mean for the future of Intel?
$1bn Fund to Promote RISC-V & Other Start-ups
Recently, Intel announced that it will be launching a $1bn fund targeted at customers of their latest foundry service called “Intel Foundry Services”. The primary goal behind the fund is to help companies accelerate their projects and development cycles that will use Intel’s foundries to produce the next generation of devices, including those based on x86, ARM, and RISC-V.
Help via the fund will include access to intellectual property, software tools, new chip architectures, and advanced packaging technologies. The foundry service will also allow Intel to offer new packaging technologies such as die-to-die modules that incorporate multiple dies stacked on a single large interconnect die.
Furthermore, Intel is also pushing for an open interconnect standard that will allow dies to communicate with each other at high speed. While there are already bus protocols (such as PCIe), die-to-die communications are high-speed, and thus a new protocol based on this would allow for interconnected dies to behave as a single die. Therefore, future designs could incorporate an Intel CPU core for high-speed processing while multiple RISC-V cores can be used for offloading I/O blocking tasks.
Why is this move a major shift in tactic for Intel?
Intel was the first company to introduce an entire CPU on a single chip, the 4004 and has been the market leader for CPUs ever since. Not only has Intel historically been the first to market their devices, but they also have been at the forefront of semiconductor manufacture, and these manufacturing capabilities have only ever been available to Intel for Intel products.
However, multiple failures by Intel to break several process nodes, including 7nm, saw its competitors gain ahead of the CPU race. Furthermore, the repeated delays seen in the 7nm node demonstrated to Intel that having CPU architecture rely on node technology means that any delay in the development of a node will automatically hold back CPU manufacturing capabilities.
Intel has recovered somewhat from the delays in 7nm, but other chip architectures such as ARM and RISC-V are starting to encroach in markets dominated by Intel, which could spell disaster for Intel long-term. However, even if Intel stopped designing CPUs, semiconductors are still needed. Intel has finally realised that opening its foundry services to others would provide an additional revenue stream for Intel and provide it with the opportunity to offer its wide range of intellectual property to customers, such as die stacking die interconnects.
Thus, Intel may now start to focus more on providing custom manufacturing services as opposed to the design and manufacture of CPUs, essentially becoming the new TSMC.
What could this mean for the future of Intel?
There is no doubt whatsoever that this move by Intel is the best thing to do; Intel excels at semiconductor manufacture, holds a wide range of patents around semiconductor design, and has the infrastructure to start supporting external customers. Furthermore, other CPU architectures are growing, and it won’t be long before x86/x64 is replaced on the grounds that it is decades old.
Intel can still participate in CPU design, but its focus may shift from consumer devices to high-end computing markets such as supercomputers. ARM and RISC-V can be used in these environments. Still, it is tough to beat a supercomputer based on a CISC architecture that is primarily designed to handle complex operations (assuming equal core counts). That is not to say that ARM or RISC-V cores won’t be used in supercomputers (as ARM already is), but Intel CPUs are well known for being the Ferrari of computational power.
Overall, the move to start manufacturing customer devices is not only sensible, but ideal for long-term growth. Semiconductors will continue to become more important, their demand will grow, and Intel’s expertise in this area will ensure its survival.