Intel's new CEO Lip-Bu Tan has once again overturned last quarter's strategy, making another major adjustment to the controversial Intel Foundry business. Whether this massive ship can set sail again remains to be seen, but the market has already cast a vote of confidence.
Intel Q3 Earnings:
Revenue was $13.65B, up 3% year over year and 6% sequentially, slightly above the prior guidance range of $12.6-13.6B. Q4 revenue midpoint guided to $13.3B, down 7% year over year and 3% sequentially.
GAAP gross margin was 38.2%, up 23.2 percentage points year over year and 10.7 percentage points sequentially. Non-GAAP gross margin was 40%, up 22 percentage points year over year and 10.3 percentage points sequentially. Q4 Non-GAAP gross margin guided to 36.5%, down 3.5 percentage points sequentially.
GAAP operating income was $680M, swinging to profit year over year and sequentially. Non-GAAP operating income was $1.52B, swinging to profit year over year and sequentially. Q4 Non-GAAP operating income guided to $320M, down 77% year over year and 79% sequentially.
GAAP net income was $4.06B, ending six consecutive quarters of losses, primarily driven by gains from the Altera stake disposal and a $3.77B net contribution from US government custodial shares. Q4 guided to a net loss of $630M.
Non-GAAP net income was $1.02B, swinging to profit year over year and sequentially, versus prior guidance of breakeven. Q4 Non-GAAP net income guided to $360M, down 36% year over year and 65% sequentially.
Q3 operating cash flow was $2.55B. Gross capex was $3B. Net capex was $1.65B. Adjusted free cash flow was $900M, ending four consecutive quarters of negative free cash flow.
Total equity was $106.4B as of quarter end.
Q3 Segment Details:
Starting in 25Q1, Intel NEX Edge moved to CCG, NEX Network moved to DCAI, IMS moved from Intel Foundry to All Other, and Automotive moved from All Other to CCG. The result is that CCG and DCAI numbers look better.
CCG revenue was $8.54B, up 5% year over year, ending three consecutive quarters of year-over-year decline, and up 8% sequentially, representing 63% of revenue. Operating income was $2.69B, down 8% year over year, with an operating margin of 31.6% (AMD 25Q2 21%), remaining Intel's most profitable business.
Driven by five years of post-pandemic replacement demand, Win10 end-of-life enterprise upgrades, and rising AI PC penetration, demand for legacy-node products is very strong. AI PC shipments grew double digits sequentially this quarter. The company expects to deliver approximately 100M AI PCs for the full year. 2026 PC TAM is expected to continue growing. Intel acknowledged a significant gap versus AMD in the high-end desktop market that needs to be closed, but stated its position in consumer and commercial notebooks remains very strong.
DCAI revenue was $4.12B, down 1% year over year and up 5% sequentially, representing 30% of revenue. Operating income was $960M, up 153% year over year and 52% sequentially, with an operating margin of 23%, rising for the fourth consecutive quarter and the highest since Q2 2022. Management emphasized that Intel server CPUs remain the preferred choice for AI head nodes, with strong Granite Rapids demand.
Intel also stated it will establish an ASIC design services business unit to provide custom chips for a broad range of external customers, extending the reach of its core x86 IP. On AI GPUs, management positions Intel as the preferred compute platform for AI inference, planning annual cadence of inference-optimized GPU iterations with enhanced memory and bandwidth to meet enterprise needs. Collaboration with NVIDIA is expected to raise the TAM.
Intel Foundry revenue was $4.24B, down 2% year over year and 4% sequentially. Operating loss was $2.32B, narrowing year over year and sequentially. Operating loss rate was 55%.
Last quarter, management stated that Intel 18A and 18AP remain the key nodes for the next three generations of products, and that internal products alone could yield a reasonable return on 18A investment, not ruling out the possibility of no external customers. Intel 14A would be a foundry node from the start (2028/2029), but investment would only occur once orders materialized. The market viewed these two points as a major shift in foundry strategic priority. This quarter, management did a 180-degree reversal.
Recently, Intel announced major progress on its 18A process, with yields improving continuously, and claimed the first 18A-based notebook CPU, Panther Lake, will ship in 2025 with volume availability in early 2026. Especially after the US government took a stake in Intel, there are requirements on its Foundry ownership stake, and Intel cannot relinquish control of its own Foundry business.
Management stated this quarter that Intel 18A yields are continuously improving but still need a full year of refinement to reach appropriate gross margin levels. Intel 18AP development is progressing. Overall, 18A capacity will not reach full utilization until 2030. The Intel 18A series will be the foundation for at least the next three generations of Intel PC and server products. Intel 14A remains in active engagement with potential external customers. Management is encouraged by early feedback, with 14A showing better performance and yield at the same stage than prior nodes. Advanced packaging continues to progress well. Foundry investment will be prudent, expanding capacity only when external customer demand commitments are secured.
Mobileye revenue was $500M, up 4% year over year, growing for the third consecutive quarter, down slightly sequentially. The revenue gap with Qualcomm has widened further. Operating income was $74M, down 5% year over year. Average system price was $51.7, down 3% year over year. Intel continues to monetize Mobileye shares.
Altera revenue was $386M, down 7% year over year. Operating income was $100M. Altera will be deconsolidated in Q4.
Call Highlights:
Management guides Q4 revenue of $12.8-13.8B. At the midpoint, this implies a 7% year-over-year decline and 3% sequential decline. The $400M FPGA business Altera will be deconsolidated in Q4. Excluding Altera impact, revenue would be roughly flat sequentially.
Q4 CCG revenue is expected to decline sequentially. DCAI revenue is expected to grow strongly sequentially. Wafer capacity will prioritize the server business. Foundry revenue is expected to grow sequentially.
Q4 Non-GAAP gross margin guided to 36.5%. 2026 capex guided to $16B (unchanged). No longer providing 2026 gross margin guidance (previously guided 40-60% Non-GAAP gross margin for 2026). Altera deconsolidation and Lunar Lake shipments in the first half of next year will continue to pressure gross margin, but Foundry gross margin will certainly improve.
2025 gross capex guided to $18B (unchanged). 2026 gross capex guided to approximately $18B (previously $9-18B). This impacts semiconductor equipment vendors such as ASML.
Intel 10/7 process capacity limited the ability to fully meet data center and PC product demand in Q3. Intel will not expand 10/7 process capacity. Substrate is also in short supply, relying primarily on inventory. The capacity shortage peak is expected in Q1 next year.
Q3 received $5.7B from the US government, $2B from SoftBank Group, $4.3B from the Altera transaction close, and $900M from Mobileye share sales. NVIDIA's $5B investment is expected to close by end of Q4.
Previously noted at the end of the last earnings report:
After overturning the IDM 2.0 strategy last quarter, this quarter Lip-Bu Tan did a 180-degree reversal and revived Pat's IDM 2.0 foundry dream. With capital from various heavyweight investors and positive 18A progress, Intel's stock has surged in the short term, completely breaking free from its previous valuation framework.
There are now two flag-waving camps on whether Intel's market cap is reasonable: if you don't believe 18A can stage an against-the-odds comeback, current valuation is indeed too high. But if you do believe in 18A, then a Foundry business losing $10B a year, once revitalized by 18A volume, can unlock substantial profit.
