Intel Foundry, the company's chip manufacturing business, had $7 billion in operating losses for 2023 and won’t break even until at least 2027, says CEO.
Intel’s accounting separation between its foundry (chip manufacturing) operations and its chip sales business shows that there’s still more money to be made in selling the chips than in making them, even after several years of declining PC sales
Intel reported overall revenue down 14 percent year on year to $54.2 billion for its fiscal year ending Dec. 30, 2023. Net income totaled $1.7 billion, most of it from investment gains, interest, and tax refunds, with operating income accounting for just $93 million.
But that small figure conceals two wildly different outcomes for Intel’s chip manufacturing and chip sales businesses, revealed by the company’s publication of its new financial reporting structure on Tuesday.
Operating income from Intel Products (including chips for laptops, desktops, data centers, AI applications, networks, and edge computing) totaled $11.3 billion on revenue of $47.7 billion.
On the other hand, Intel Foundry, the division that makes many of Intel’s own chips as well as some for other companies, made an operating loss of $7 billion, up from $5.2 billion a year earlier, while its revenue declined 31 percent to $18.9 billion.
More losses to come
Intel CEO Pat Gelsinger was frank Tuesday in statements to investors and analysts that 2024 would continue to see significant losses in the foundry business, which likely won’t break even until at least 2027.
“2024 is the trough for foundry operating losses,” he said in comments to investors after Intel’s earnings announcement Tuesday. “We’ve committed to being the number-two foundry by the end of the decade. And between now and then, we’ll hit breakeven operating margin about midway through that and then driving operating margin improvement through the period.”
Gelsinger cited two major factors for Intel’s continued difficulties: the decline in sales of PCs using Intel chips following the boost in purchases during the COVID-19 pandemic, and the company’s small share of the market for chips to accelerate the training of AI models.
“We also expected that we would have been more successful with our accelerator share gains and where we would be by this point in time,” he said. “So those would be, I say, the two biggest headwinds that weren’t accounted for before.”
Cautious optimism on future of foundry
However, Gelsinger was cautiously optimistic about the foundation Intel is building to drive manufacturing earnings growth as it repositions its foundry business for manufacturing not only its own chips, but also chips for other companies.
“We’ve tried to take very reasonable, modest assumptions for our core business where we’re sort of in the core area, in the mid- to low single digits growth rate,” he told investors and analysts. “Additionally, some improvements or additional revenue growth for our foundry business that we’re seeing as we look over that cycle as well, leading us to the mid- to upper single digits for an aggregate growth rate in the modeling that we’ve seen here.”
Intel’s long-time chip rival AMD spun off its chip-manufacturing business as GlobalFoundries in 2009, and sold the last of its stake in 2012.
Rather than spin off its foundry unit, though, Intel decided in 2021 to set up its manufacturing and processes business as a separate foundry unit within the company as part of its Integrated Device Manufacturing (IDM) 2.0 plan.
That was a good decision, noted Gaurav Gupta, vice president analyst in Gartner Group’s emerging tends and technologies team, and should help drive Intel’s future margins as it gains efficiencies in this business, he said.
The move also “opens options for product groups developing CPUs for various applications to have an open choice of foundry and to be more efficient in terms of their test/wafer stepping requirements,” Gupta told Computerworld.
Pareekh Jain, CEO of Pareekh Consulting, agreed. Intel’s foundry plan is good, but its success will depend upon Intel’s execution, especially how early the company can build new capacity and improve utilization.
“Intel’s foundry margin will improve with technology improvement, increasing utilization and capital and cost efficiency in new fabs,” Jain told Computerworld.
Conflicting strategies?
Still, even as Intel builds out its foundry business, it will continue to outsource the manufacturing of some chips — either for cost/performance reasons or just the fact that Intel doesn’t fabricate everything needed for its products, Gupta noted.
This could lead to conflict with its ultimate plan to increase its fabrication footprint and renew its focus on manufacturing to reduce dependence on external foundries and thus regain customer confidence, he said.
“One of the current concerns in outsourcing to external foundries is that it shows lack of trust in its own manufacturing, which some of Intel’s potential foundry customers are questioning,” Gupta observed.
Indeed, Intel’s continued use of external foundry partners in Taiwan “contradicts Intel’s stance of developing a domestic resilient supply chain for fabless customers in the US,” he said, particularly since the chipmaker “is receiving top dollars from the CHIPS Act” to do so.
