Intel wants to become like TSMC when it comes to manufacturing chips. TSMC is the world’s largest contract manufacturer for integrated circuits. With it, all you need to do is to design your circuit, send to it and it will manufacture it, and send the chips back to you. You can call it Amazon AWS as in cloud computing where the cloud infrastructure which anyone can rent with a debit or credit card is equivalent to foundries worth $billions which TSMC has built to turn sands into smart systems for the world.
So, just as cloud computing has done well, TSMC has also done really very well. And Intel wants to copy that business model as it looks for ways to get out of the paralysis it has found itself. So, Intel is opening its foundries for friends and enemies, provided you are open to pay: “Buffeted by the success of Samsung, TSMC, and Apple’s new M1 chip, Intel is announcing a new business strategy to compete in the next era of semiconductors. It will start to make chips for Qualcomm, one of the world’s biggest chip designers — the first time it’s opened its foundry to a rival.”
Earlier this year, Intel got a new CEO and kicked off a new business plan that would open its foundries to other chip-design firms, the same way TSMC and Samsung Semiconductor operate. At its “Intel Accelerated” event today, the company laid out a roadmap for its future as a for-hire foundry. Besides the future of ever-smaller process nodes, the company also announced it has scored one of the world’s biggest chip designers, Qualcomm, as a future foundry customer.
As part of entering the foundry market, Intel will start naming its process nodes more like its rivals. The process-node numbers used for chips like “5nm” started out life as a measurement of transistor size, but eventually the marketers got hold of them and companies started cheating down their numbers to look more advanced. Intel says its new naming scheme will better align with how TSMC and Samsung talk about their foundry technologies. Gone are the days of “Intel 10nm Enhanced Super Fin”—instead, the node is called “Intel 7.” It should have a comparable density to the TSMC and Samsung 7 nm nodes and will be ready for production in Q1 2022 (TSMC and Samsung are currently shipping “5nm” products). “Intel 4″—which Intel previously called “7nm”—is now said to be equivalent to TSMC and Samsung’s 4 nm node, and it will begin manufacturing products in 2023.
This is a very great strategy, and statistically it makes sense for Intel. I have written on this many times and my argument is like this: building foundries costs $billions of dollars making the business risky when Intel has to fab for only its products. If it does not have a product winner in the market, the foundry will become highly underutilized. That risk of usage efficiency affects its capacity to modernize its factories, and because of that inertia, it has lost ground to Samsung and TSMC which fab for anyone that can pay.
Samsung serves Apple while TSMC serves everyone. For those companies, they do not have to depend on internal design wins to fill foundry capacities since those winners could come from any of their many customers. If Samsung Galaxy does not do well in the market but Apple iPhone does well, Samsung Electronics will be fine since it helps Apple to manufacture the core microprocessor in iPhone. That is at the heart of the One Oasis Strategy which I wrote in Harvard a few months ago.
With that confidence that even outside clients can have the winner, companies like TSMC and Samsung can invest in new fabs since statistically one of their many customers will win the next evolution. But the old Intel was tethered to its internal success which is not always guaranteed. If a processor series does not do well, the foundry investment will struggle.
But today, this new strategy will unlock massive opportunities in Intel and I think it is a great win. Intel will serve Amazon AWS, Qualcomm and who knows companies like Nvidia and AMD in the near future, freeing the foundry from pure dependence on its internal designs. If you look at it, that is a massive evolution for the semiconductor giant. And it is a great one.
Comment: I’d like to view this from the point of national security. If conflict starts around Taiwan today and very close to Asia, the US will be cut off from critical and cutting edge manufacturing capabilities. The China – Taiwan – US feud isn’t going anywhere soon. There is need for a backup.
My Response: I made that point here but note that TSMC is building a factory in US to deal with that – https://www.tekedia.com/tmsc-to-build-a-12bn-semiconductor-factory-in-the-u-s/
Comment: It was always clear that Intel has to start thinking like a startup to be competitive in today’s fast paced semiconductor industry. Closed source business models will not survive in the next couple of years and the early effects are what we can clearly witness with Intel losing relevance in this sector.
The cost of computing power is already past Moore’s law where all manufacturing has fast caught up with design metrics with more and more people having access to latest tech. The producers rather than then architects are clearly winning in this regard. Collaboration is Intel’s best bet in surviving the onslaught of manufacturing over architecturing.
More and more open source chip architectures such as RISC-V will play a huge role stealing market share from the big players in the years to come.
Comment: In the long run nations, businesses and organizations must learn to live, interoperate and interact with rivals to keep the numbers good, to remain relevant, the current world architecture calls for more openness than reclusiveness, we must embrace the challenge while we build requisite capacity to compete.---
1. Advance your career with Tekedia Mini-MBA (Feb 7 – May 7, 2022): 140 global faculty, online, self-paced, $140 (or N60,000 naira). Click and register here.
2. Click to join Tekedia Capital Syndicate and own a piece of Africa’s finest startups with a minimum of $10,000 investment.