Intel has entered the recovery phase • The Register

comment Intel's stock price rose sharply on Monday as it was announced that the company was spinning off its foundry business into an independent subsidiary and adding AWS and the Department of Defense as customers.

But while Wall Street celebrated, Chipzilla's road to recovery is far from over, and everyone involved – shareholders, employees, and partners – will have to grit their teeth and endure either until CEO Pat Gelsinger can realize his vision or until the board ends its term early.

Some are already feeling this more than others. Along with the spin-off, Gelsinger announced sweeping changes to the company's structure. The network and automotive groups will be merged with the customer division. And – perhaps even more worrying – the development of the German manufacturing plants and Polish assembly sites will be suspended for two years.

The announcement has raised doubts about the future of these facilities, not to mention the EU's goal of doubling its share in semiconductor development, manufacturing and material supply from 10 to 20 percent by 2030.

EU member states have already provided 43 billion euros ($48 billion) in subsidies to support that goal. Intel was originally set to receive around 12 billion euros ($13.4 billion) in state aid for its German and Polish developments. The aid is unlikely to ever materialize, and without it, Intel's two-year delay could be permanent.

The timing of the announcement was particularly unfortunate, as it came just days after the Polish government received the green light from the European Commission to finance Intel's planned test and assembly facility.

But even with the financing, these facilities would have been extremely expensive. Intel would have had to raise more than 20 billion euros (22.3 billion dollars) in capital – and this at a time when the company is drastically cutting costs and personnel. By the end of 2024, Intel will have almost 16,000 fewer employees worldwide.

Intel has been one of the few leading chipmakers, and unless someone else fills the gap, Europe may soon be reduced to producing little more than museum-quality chips.

Samsung is undoubtedly big enough to enter this market, but even with a few billion in subsidies to mitigate the negative aspects, it may not be economically attractive enough to do business in Europe.

Intel's Foundry spin-off won't change much

While Intel is clearly prioritizing its U.S. investments, despite all the fanfare, the spinoff won't stop Foundry from bleeding, even as AWS and the Department of Defense line up to get their share of the capacity.

At the same time as the spin-off, AWS confirmed plans to manufacture an AI fabric chip and announced it would commission a custom version of the x86 giant's upcoming Xeon 6 processors. Meanwhile, Uncle Sam has pledged $3 billion to Intel as part of a program called “Secure Enclave” to build a secure semiconductor supply chain.

While it's a much-needed win for Intel Foundry, it will be some time before Intel can generate revenue from either deal. The fact is that Intel's foundry business has struggled to turn a profit — the group reported operating losses of $2.8 billion last quarter — and that's not likely to change anytime soon as an independent subsidiary.

So what exactly is being spun off? From what we hear, not much. Under the new structure, Intel Foundry will operate as an independent subsidiary within Intel. That means the company will have its own board of directors, more autonomy in finding new sources of funding, and potentially a better reputation among fabless chip companies that are wary of building chips for a rival company.

But even compared to Intel's spin-off of Altera late last year, Gelsinger is keeping a tight grip on the division. Rather than electing a new CEO to lead the company, Intel Foundry employees will continue to report to him.

This will no doubt change over time, and we expect an IPO to be on the horizon, but for now, Intel's product future is closely tied to Foundry's success. With the decision to move its Arrow Lake CPUs to TSMC, very little of Intel's 2024 product portfolio will be manufactured in-house. However, if all goes to plan, this will change when production of Intel's 18A process node starts in 2025 and reaches volume production in 2026.

It is imperative that Intel implements this plan without delay. Unlike TSMC, which is supported by leading customers such as Apple, Nvidia and AMD, Intel remains Foundry's largest and most important customer.

It won't be easy to give up Gelsinger's dream

When Gelsinger returned to Intel as CEO in early 2021, he immediately laid out an ambitious plan to reshape the struggling chipmaker.

This “all or nothing” strategy was reflected when, less than a month after taking office, he surprised many with the announcement that he wanted to open Intel's factories to contract manufacturing and invest $20 billion – a sum that has since grown to around $30 billion – in two new state-of-the-art factories in Arizona.

In the years that followed, Intel announced the development of a new “angstrom-era” process technology, committed to building two more factories in Ohio worth about $20 billion, enlisted the help of private equity firms such as Brookfield and Apollo to finance the projects, and secured $11.5 billion in government grants under the CHIPS and Science Act.

This made it incredibly difficult for Gelsinger to give up his foundry division. Brookfield and Apollo have invested around $26 billion in the success of Intel Foundry. Uncle Sam is now even more dependent on Intel.

After GlobalFoundries abandoned development of its 7nm process node in 2018, Intel became the only domestic supplier of cutting-edge process technology in the United States, making the company arguably the country's most important chipmaker and closely intertwined with American national security policy.

To be clear, the United States cannot afford to let Intel Foundry fail, especially given China's successes despite increasingly stringent trade restrictions.

And even if Intel tried to sell Foundry, who would be crazy enough to buy a company that makes billions in losses every quarter?

Gelsinger sees this venture as Intel's biggest transformation since moving from memory to microprocessors. That may be true, but nobody said it would be easy…or fun. ®