Tower Semi's $1.3B prepaid SiPho contracts de-risk TSEM's 2027 revenue. STM's new sensors cut power 10x. Alpha Score 55 and 54. Next catalysts ahead.
On May 13, Tower Semiconductor Ltd. (TSEM) signed Silicon Photonics (SiPho) customer contracts valued at $1.3 billion for 2027 revenue. The company has already collected $290 million in customer prepayments for capacity reservations from its largest clients. That initial commitment is backed by an even larger contractual wafer commitment for 2028, with corresponding prepayments due by January 2027.
The capacity reservations represent minimum contractual commitments. Total shipment forecasts and demand from over 50 active SiPho clients are expected to be higher. To meet this demand, Tower is executing a global multi‑fab capacity ramp. This expansion underpins its 2028 long‑term financial model, which targets $2.8 billion in total revenue and $750 million in net profit.
Bottom line for traders: The prepaid structure forces customers to commit capital early, reducing TSEM’s revenue uncertainty before a single chip ships. A missed prepayment deadline would signal real demand weakness.
Multi‑year agreements are driven by rapid AI infrastructure growth and escalating data center bandwidth demands. Tower is investing in advanced optical architectures, including 400 GHz/lane performance, optical circuit switches, and 3DIC hybrid‑bonding. The company is also collaborating with industry innovators on next‑generation modulators for ultra‑high‑bandwidth data transmission.
Tower is a leading independent foundry specializing in high‑value analog semiconductor solutions. It manufactures customized analog ICs – including radio frequency (RF), power management, and CMOS image sensors – for over 300 customers.
On April 28, STMicroelectronics (STM) introduced its next generation of ultralow‑power global‑shutter image sensors: the monochrome VD55G4 and the RGB color VD65G4. Part of the ST BrightSense portfolio, these compact, microcontroller‑compatible sensors are engineered for battery‑operated or energy‑harvesting personal electronics. They target high‑growth applications including wearables, AR/VR/XR headsets, smart home appliances, and portable medical devices.
The new sensors feature an optimized detect‑and‑wake architecture that allows them to consume up to 10x less power than conventional global‑shutter sensors during standard operation. By continuously monitoring a scene and waking the main processor only when changes occur, they enable event‑driven operation. This reduces standby power and extends battery life while delivering AI‑ready data locally at the edge.
Manufactured on 300 mm wafers using a 3D‑stacked 65 nm / 40 nm architecture at STMicroelectronics’ Crolles plant, the sensors are currently available to early adopters. To accelerate product development, STM is rolling out a comprehensive companion ecosystem including turnkey camera modules, evaluation software, and dedicated development boards for popular platforms like STM32 and Raspberry Pi.
STMicroelectronics operates through four segments: AM&S (Analog products, MEMS & Sensors Group), P&D (Power and Discrete products), EMP (Embedded Processing), and D&RF (RF Products). The new sensors sit squarely in the AM&S segment, which benefits from the shift to edge AI. Wearables and AR/VR headsets require sensors that can process data locally without draining batteries. STM’s 10x power reduction directly addresses that bottleneck.
Key insight: The sensor launch does not carry the same revenue visibility as Tower’s prepaid contracts. A design win from a major OEM – likely in wearable or smart home space – would be the first concrete proof point.
Deloitte projects 2026 global semiconductor sales will hit a historic peak of $975 billion, driven by a 26% growth rate fueled by AI infrastructure. The boom masks a stark structural paradox: high‑value GenAI chips are expected to generate roughly half of the industry’s total revenue ($500 billion) while making up less than 0.2% of total unit volume.
This value concentration has propelled the market capitalization of the top ten chip companies to $9.5 trillion, with the top three dominating 80% of that total. Meanwhile, non‑data center segments like PCs and smartphones face slower growth and projected declines due to soaring memory prices. While 2026 revenues are secured by backlogs, the industry faces sharp risks for 2027 and 2028 if organizations experience delayed AI return on investment, severe power grid constraints, or deflationary pricing from new competitors.
On May 14, Ankur Crawford, Alger EVP, told CNBC that she is not worried about a potential market bubble forming. She explained that earnings numbers are continuously rising for all AI‑related stocks and that these numbers are still too low relative to their actual potential for the next two to three years.
Crawford recalled appearing on the show in 2024 and predicting that CapEx would rise sequentially through 2025, 2026, 2027, and 2028. For Tower and STM, the risk is concentrated in the non‑AI portions of their businesses. Tower’s SiPho contracts are tied directly to AI data center buildout. STM’s sensor play is more diversified across consumer and industrial end markets, which could face slower growth if the AI capex cycle falters.
AlphaScala’s proprietary scoring system gives TSEM an Alpha Score of 54/100 (label: Mixed) and STM an Alpha Score of 55/100 (label: Moderate). Both sit in the Technology sector. The scores reflect the companies’ current positioning. TSEM benefits from a clear, multi‑year contract backlog but faces execution risk in ramping SiPho capacity. STM has a broader product portfolio but lower growth visibility outside of sensors.
| Metric | Tower Semiconductor (TSEM) | STMicroelectronics (STM) |
|---|---|---|
| 2028 Revenue Target | $2.8 billion | Not disclosed |
| 2028 Net Profit Target | $750 million | Not disclosed |
| Alpha Score | 54/100 (Mixed) | 55/100 (Moderate) |
| YTD Performance | >130% | >130% |
| Key Catalyst | $1.3B SiPho contracts | Ultralow‑power sensors |
For TSEM, the next concrete marker is the January 2027 prepayment deadline for the 2028 wafer commitments. If those prepayments come in on schedule, it confirms that customer demand is real and that the 2028 revenue target is achievable. A delay or reduction would signal that AI infrastructure spending is slowing.
For STM, the catalyst path depends on adoption of the VD55G4 and VD65G4 in wearable and AR/VR designs. Early adopter shipments are already underway. The next catalyst is a design win announcement from a major OEM – likely in the wearable or smart home space. Without such a win, the sensor launch remains a product announcement without revenue impact.
Both stocks trade at elevated multiples relative to historical averages. The prepaid contract structure for TSEM and the power‑efficiency moat for STM provide tangible buffers. The simple read is that both are riding the AI wave. The better market read is that TSEM has de‑risked its 2027 revenue through customer prepayments, while STM is betting that edge AI will demand sensors that consume an order of magnitude less power. The next six months will determine whether those bets pay off.
For a broader perspective on semiconductor stock performance and valuation, see our stock market analysis. Access individual profiles for TSEM and STM.
Prepared with AlphaScala research tooling and grounded in primary market data: live prices, fundamentals, SEC filings, hedge-fund holdings, and insider activity. Each story is checked against AlphaScala publishing rules before release. Educational coverage, not personalized advice.