Intel’s Foundry Grew 31% Last Quarter and Lost $2.1 Billion Doing It.
Intel (INTC -4.06%) posted its fastest revenue growth in nearly 15 years last month, with second-quarter revenue rising 25% year over year to $16.1 billion. But the unit the company’s whole transformation is staked on, Intel Foundry, is still deep in the red. The chipmaking arm lost $2.1 billion in the quarter on $5.8 billion of revenue.
The loss, I’d argue, is where the progress lives. A year ago, the foundry lost $3.2 billion on $4.4 billion of revenue — about 72 cents lost for every dollar the unit brought in. This quarter it lost about 36 cents per dollar. The loss per dollar of revenue halved in a year.
Intel has told investors the foundry should stop losing money in 2027, and the growth stock (up more than 160% in 2026 as of this writing) is priced as if that arrival is on schedule. So how much revenue does the foundry need before the losses stop?
Image source: Intel.
A smaller loss on more revenue
The trend is now three quarters deep and pointed one direction. Intel Foundry’s revenue climbed from $4.4 billion in the year-ago quarter to $5.4 billion in the first quarter of 2026 and $5.8 billion in the second. Its operating loss, meanwhile, narrowed from $3.2 billion to $2.4 billion and then $2.1 billion. And the second quarter’s 31% year-over-year revenue growth was itself an acceleration, up from 16% in the first quarter.
The company’s explanation is about the factories themselves. Production on Intel 18A, the company’s newest widely deployed manufacturing process, came in about 25% above Intel’s internal target and rose more than 50% from the first quarter. Better yields and faster cycle times are bringing down wafer costs, and Intel says the foundry has cut the cost of its main Panther Lake chip by roughly 50% so far this year.
Clearly, the factories are getting cheaper to run.
Intel’s biggest customer is Intel
But external customers supplied just $293 million of the foundry’s $5.8 billion in quarterly revenue. The rest (roughly 19 of every 20 dollars) came from building chips for Intel’s own product groups.
Of course, that internal demand is no small thing. Intel’s data center and AI segment grew revenue 59% year over year last quarter, and the company says it is supply constrained, with data center customers demanding more chips than Intel can produce.
And chief financial officer David Zinsner said in the company’s published earnings call remarks that customers “continue to signal a strong and sustainable spending environment.” After all, a foundry filled by its owner’s orders beats an empty one.
But internal revenue can only prove the factories work. It can’t prove the business does.
Intel still hasn’t announced a major outside customer for its leading-edge processes — security specialist Fortinet, which in July became the foundry’s first named customer under CEO Lip-Bu Tan, is buying chips built on an older process. Until other companies’ chips fill these factories at scale, the foundry rises and falls with Intel’s own product cycle.
Today’s Change
(-4.06%) $-4.13
Current Price
$97.52
Key Data Points
Market Cap
Day’s Range
$96.30 – $100.03
52wk Range
$20.76 – $142.35
Volume
101.2M
Avg Vol
123.1M
Gross Margin
39.05%
Break-even has a date
Intel‘s stated target is foundry break-even in 2027. Zinsner said last year that getting there requires only a few billion dollars of additional external revenue. That’s a small number against the unit’s $23 billion annual revenue pace, and a large one against the roughly $1.2 billion annual pace external customers supply today.
The bridge is supposed to be Intel 14A, the next manufacturing process, which is being prepared for risk production — early trial manufacturing — in 2027 with high-volume output committed for 2028. Meaningful outside volume, in other words, likely arrives near the deadline, not ahead of it. And the spending comes first: Intel says it is substantially increasing its investments to support the demand it sees. On Monday, the company announced a $15 billion common stock offering to help fund the build-out without adding new debt.
To me, the external revenue line is the one to watch, and the arithmetic hasn’t changed: outside customers pay Intel about $1.2 billion a year, and break-even needs a few billion more.
The loss math is improving at a pace that, I think, makes 2027 believable. The stock is another matter. At about $98 as of this writing, shares trade near 60 times next year’s expected earnings — a rich valuation by any standard, and a price that assumes the foundry bet has already been won.
The foundry is doing what Intel said it would, a quarter at a time. At this price, the stock needs it to keep doing exactly that for two more years.
