More to come, raise TP to $54: On Nov 28, Intel share price was up 10% owing to news of order win for Apple’s M-series in 2027. We’ve highlighted the potential order win of M-series in our earlier Intel report (Belief in the Story Takes Hold on Oct 24) and indeed, the progress is earlier than expected. We expect more to come and now expect the upcoming win in Apple (AAPL Buy) non-pro smartphone SoC based on Intel 14A for 2028; coupled with the likely engagement with Nvidia (NVDA Buy) and AMD (Buy)’s server SKUs, we believe its 14A will be a strong node. For back-end, we expect Intel to benefit from the CoWoS undersupply and potential “Made in USA” trend. As highlighted in our MediaTek (2454 TT Buy) Upgrade report on Nov 28, we expect TPU v8e to adopt EMIB, and we also expect Apple to be an important customer. Near term, despite memory concerns, server CPU sales are expected to accelerate due to increasing general purpose compute. In all, we believe the recent developments bode well for IFS in the longer term. We raise TP from $50 to $54, now based on 2x FY27E BVPS, reflecting the better visibility of earnings improvement trajectory.
18A the key to build customers’ confidence: According to Intel, Panther Lake is on track to enter HVM and debut on Jan 5 at CES 2026. We believe the yield rate has reached 60-65% in November and targets 70% by end-2025. We also expect the entry-level Wildcat Lake to start HVM in 1Q26, with much higher yield rates due to smaller die area.
Robust IFS customer pipeline with top-tier U.S. fabless: We still expect Intel to ramp 18A and 18A-P at small scale in 2026 to ensure yield and quality. That said, the external customer is expected to start by end-2026. In addition to the aforementioned Apple M-series on 18A-P in 2027, we expect 14A to be driven by the potential wins for: 1) Apple’s non-pro smartphone SoC in 2028 (we expect firm visibility by end of 2025); 2) Nvidia & AMD server projects in 2028; 3) Tier-1 Ethernet switch in 2028. With thoughtful cost/scale management, we expect its IFS GM turnaround by end-2026, followed by a strong ramp of 14A in 2027.
Backend EMIB to win AI accelerators on “TSMC + Intel” model: We believe that EMIB emerges as an alternative to CoWoS, given the capacity constraints at TSMC (2330 TT Buy), higher design flexibility and larger reticle size (vs. OSAT’s). As said, we expect TPU v8e to adopt EMIB-T in 2027, as Google provides a very aggressive volume in 2027 and 2028. We also see high chance for Intel to win more ASICs, likely Broadcom and Apple’s. Capacity wise, we expect Intel to add equipment order in 1Q26. Aside from current Malaysia factory, it would be making sense for Intel to accelerate in New Mexico Fab to benefit from TSMC’s Arizona. In sum, we estimate its advanced packaging business will contribute >$1bn revenue in 2027E, with growth in the following years through more project wins.