Intel is not in a good place. We and everyone else has written a lot about this. Their entire model is built on having the best manufacturing process, but that side of the company has stumbled, badly. They are bleeding share in their Data Center Group (DCG) source of roughly half of their profits. They are still highly profitable, but surprisingly low in cash flow. Working in their favor is the immense amount of technical talent they have, a still solid PC CPU business and a new CEO who maybe is going to move them in the right direction.
A common them from us and many others holds that Intel has basically two options – they can either double down and solve their internal problems or outsource manufacturing to TSMC who is far ahead of them right now in manufacturing processes. Instead they have selected Option C – All of the Above. Yesterday, they announced they are going to invest $20 billion to massively expand their plant in Arizona, building two new fabs. They are going to outsource more production to outside foundries. And crucially, they set up a new division, Intel Foundry Services, to provide manufacturing for other customers.
This a bold plan, not the action of a company on the ropes. Their new CEO seems to have done a frank inventory of where the company stands and decided that options like splitting the company or selling to private equity were not necessary. Presumably he has some solid basis for that confidence.
Now the company has two challenges.
First, they have to deliver on manufacturing. All that investment will come to nothing if those new plants end up operating on trailing manufacturing processes. As far as we can tell, they should be able to accomplish this, but then again it has never been entirely clear how they went so far astray. And not for nothing, this CEO is not the first Intel CEO to promise massive investments in the Arizona desert, he is not even the second Intel CEO to do so. Nonetheless, they really should be able to accomplish this, they have such deep expertise and investment here. This could simply be a case of past CEOs taking the manufacturing side of the business for granted and under-investing in it. The new CEO appears to have a different mindset.
So really, it will be the second challenge that will really tests the company – building a customer service business for their foundries. In all those old posts we linked above, there is a common refrain. Intel has tried for close to a decade to become a Foundry, manufacturing other people’s chips. Tried and failed. The reason for this is cultural more than technical. Throughout Intel’s history, the Operations organization that runs the fabs has been the sun king around which all other parts of the company have orbited. But running a Foundry is different than running a fab, and they could never quite get the knack for a customer-is-always-right mentality. And to be clear, we are not saying the individuals involved are arrogant or uncaring, it is just a function of the whole company being organized with this planetary model.
Which is why the announcement of their Foundry Services group is noteworthy. This is an entirely separate organization whose head reports directly to the CEO, prominent and high priority. There are other forces working in their favor. When we last posted on this subject, industry expert Ben Bajarin posted his theory that Apple could make an excellent first customer for a then-hypothetical Intel foundry. That seems to be borne out by comments to the press today hinting that Apple could be on board. Intel’s announcement mentions that they have gotten very positive feedback from customers about their foundry, and the press release mentions that they will use the new capacity for Arm and RISC-V chips, not just x86..
That being said, this will not be an easy change to make. It was not clear from today’s announcement just how much influence the new Foundry org will have. Will they control one of those two plants in Arizona? How will the company prioritize customers’ orders versus internal orders? Industry leader TSMC does not face these challenges, and it is the kind of organizational problem that would bog down most companies (except maybe Amazon). And it is also not entirely clear how Intel will fund all of these, $20 billion is a lot of money, even for Intel.
It is obviously too early to know how this will shake out. At best, we can say that we are encouraged by Intel’s moves. This is a highly confident signal. Now they just have to execute on it.