By Mark LaPedus
The earnings season is here again! Here are the latest financial results from the leading foundry vendors:
Intel
It was another disastrous quarter for troubled Intel. Amid layoffs and other cost-cutting actions, Intel reported flat sales on larger losses for the second quarter of 2025.
Second-quarter revenue was $12.9 billion, flat year-over-year. Net loss was $2.9 billion in the quarter, compared to a loss of $1.6 billion a year ago.
Intel’s foundry business continues to spill red ink. The company’s troubled foundry unit reported a $3.2 billion loss on sales of $4.4 billion in the quarter. Intel has reported several consecutive quarterly losses in its foundry unit.
On a positive note, Intel reached a key milestone with the start of production wafers in Arizona based on its 18A process.
But on the downside, Intel has completed the majority of the planned headcount actions it announced last quarter to reduce its core workforce by approximately 15%. Intel plans to end the year with a core workforce of about 75,000 employees as a result of workforce reductions and attrition.
As a result of these actions, the company recognized $1.9 billion in restructuring charges in the second quarter of 2025.
As part of its cost-cutting efforts, Intel will no longer move forward with planned projects in Germany and Poland. The company also intends to consolidate its assembly and test operations in Costa Rica into its larger sites in Vietnam and Malaysia.
In addition, Intel will further slow the pace of its fab construction in Ohio to ensure spending is aligned with market demand.
Intel also recognized approximately $800 million of non-cash impairment and accelerated depreciation charges related to excess tools with no identified re-use and approximately $200 million of one-time period costs in the second quarter of 2025.
Intel is forecasting third-quarter 2025 revenue from $12.6 billion to $13.6 billion. It expects third-quarter EPS attributable to Intel of minus $0.24.
TSMC
TSMC posted mixed results for the second quarter ended June 30. The foundry giant also raised its revenue growth target outlook for 2025.
On July 17, TSMC reported sales of US$30.1 billion in the quarter, up 17.8% from the previous quarter and up 44.4% year-over-year. For the quarter, TSMC reported a net income of US$12.8 billion, up 10.2% from the previous quarter and up 60.5% year-over-year.
TSMC’s second quarter revenues were below the consensus estimate of $31.3 billion, while above the midpoint of guidance of $28.8 billion, according to KeyBanc Capital Markets.
In the second quarter, shipments of 3nm accounted for 24% of TSMC’s total wafer revenue, 5nm accounted for 36%, and 7nm accounted for 14%. Advanced technologies, defined as 7nm and below, accounted for 74% of total wafer revenue.
“Our business in the second quarter was supported by continued robust AI and HPC-related demand,” said Wendell Huang, senior vice president and chief financial officer of TSMC. “Moving into third quarter 2025, we expect our business to be supported by strong demand for our leading-edge process technologies.”
Based on the company’s current business outlook, TSMC’s third quarter sales are expected to be between US$31.8 billion and US$33.0 billion.
During a conference call, TSMC also disclosed more details about its results and talked about its future plans. Here are the main highlights:
Growth outlook
“TSMC raised its full-year target to +30% year-over-year in 2025 (from +mid-20%), driven by continued robust AI and HPC demand, especially on 3nm and 5nm nodes. Management indicated that uncertainties and risk from tariff policies have not yet caused changes in customer behavior, with no observed tariff-related pull-ins. That said, TSMC remains cautious heading into 2H25 and 2026,” according to analysts John Vinh and Steve Barger at KeyBanc, in a research note.
2nm process update
“TSMC indicated N2 is on track for volume production in 2H25, with initial demand outpacing N5/N3 ramps, driven by smartphone and HPC applications. N2P, an extension of N2 with further performance and power benefits, is scheduled to ramp in 2H26,” according to Vinh and Barger.
TSMC’s N2 process yields are currently at 65%.
Taiwan, U.S. fab updates
Over time, TSMC plans to build 11 new fabs in Taiwan. The company is ramping up one new fab in Arizona with several more U.S.-based facilities in the works.
“Arizona (AZ) fab Phase 1 (4nm) already ramped to high-volume in C4Q24, while construction of AZ Phase 2 (3nm) is already complete and TSM plans to speed up the volume production schedule by several quarters,” said Krish Sankar, an analyst at TD Cowen, in a research note. “Similarly, Phase 3 (2/1.6nm) construction has begun and TSM is also looking to speed up the production schedule.”
Packaging capacity update
TSMC’s 2.5D packaging technology is called Chip on Wafer on Substrate (CoWoS). CoWoS technology is used to assemble and stack several chips in the same package.
Demand has been strong for the technology. “CoWoS capacity remains fully loaded through 1H25, and management expects a supply-demand imbalance to persist into 2026,” according to Vinh and Barger.
But in recent times, TSMC saw some production cuts here due to Nvidia. Nvidia is ramping up the GB200 NVL72, which is basically an exascale computer in a single rack. The GB200 NVL72 connects 36 Grace CPUs and 72 Blackwell GPUs in a rack-scale, liquid-cooled design. It boasts a 72-GPU NVLink domain that acts as a single, massive GPU and delivers 30X faster real-time trillion-parameter large language model (LLM) inference.
Nvidia’s GPU architecture also includes high bandwidth memory (HBM). The GPUs and HBMs are incorporated in TSMC’s 2.5D CoWoS technology.
For some time, though, Nvidia has been struggling with the production yields with the GB200 NVL72. This in turn impacts TSMC’s CoWoS shipments.
“For the second time this year, NVDA is cutting its supply of CoWoS interposers by 5% to 365K interposers, which will support a GPU supply of ~5M units, which is slightly lower than our prior estimate of 5.5M GPUs,” KeyBanc’s Vinh said in a recent research note. “Attribution for these latest cuts is due to struggles related to the ramp of GB200 NVL72 rack production at server ODMs. However, rack production yields more recently have improved to 75%+ from 70% previously but are still below targeted hardware EMS manufacturing yields of 95%.”