STM (STM) Q3 2026 Earnings Call Transcript

STMicroelectronics reported Q3 2026 revenue of $3.76 billion, above the midpoint of its outlook, and non-U.S. GAAP EPS of $0.37; gross margin rose 410 basis points year over year to 37.3%. Management forecast Q4 revenue of $4.15 billion at the midpoint and full-year 2026 revenue of about $14.5 billion, up 22.8%, citing AI datacenter demand, where revenue is expected to reach about $1.3 billion in 2026 and well above $2 billion in 2027. Backlog exceeded $20 billion, while management expects more than 400 basis points of structural gross-margin improvement from manufacturing reshaping by late 2027 or early 2028; input-cost increases and continued consumer weakness remain risks.

Analysis

The upside is increasingly concentrated in datacenter connectivity and industrial recovery, while the earnings-quality test is whether that growth converts into factory utilization and segment profitability. SiPho capacity expansion and strong order indications create operating leverage, but also execution risk: reported customer yields and demand are management claims, and a capacity ramp does not guarantee shipments at attractive pricing. Broadcom and Marvell are relevant competitive checks in datacenter connectivity; alternatives or customer redesigns could limit ST’s pricing power. The NVIDIA robotics demonstration is not evidence of a material revenue relationship for NVIDIA.

The less appreciated drag is the gap between revenue growth and economics in Power & Discrete, which remains deeply unprofitable at the segment level, alongside pressure in Analog/MEMS. Thus, headline growth can coexist with subdued consolidated margin progress. The promised >400 bp structural uplift is a late-2027/early-2028 opportunity, contingent on fab closures and customer qualifications—not a near-term earnings reset. Input-cost inflation may absorb some mix and pricing gains. Backlog and book-to-bill can also overstate durable demand if customers are securing scarce capacity or double-ordering.

Near term, the guidance and AI/industrial momentum support relative strength, but avoid underwriting the long-dated margin target before evidence of conversion. The key catalysts are January 2027 detail on datacenter revenue and seasonality, plus quarterly evidence of SiPho shipments, utilization, and segment margins. Thesis weakens if bookings normalize toward 1, the >$2 billion 2027 datacenter ambition is cut, SiPho capacity fails to ramp, or fab qualification/closure timing slips.

Key Decisions for Investors

• Trade: consider a starter long in STM on post-earnings volatility rather than chasing the initial move; add only after January 2027 guidance confirms datacenter growth and the next results show improving utilization or margins. No price target is supportable from the supplied data.
• Keep the position sized as a cyclical growth exposure, not as a near-term 400 bp margin-recovery story. Reassess if bookings deteriorate materially, Power & Discrete remains deeply loss-making despite higher revenue, or the fab transition timeline slips.
• Monitor quarterly datacenter revenue, SiPho capacity utilization/yields, customer concentration and pricing, inventory days, and input-cost commentary. These determine whether backlog becomes profitable shipments or merely capacity commitments.
• Do not use the NVIDIA robotics collaboration as an NVDA earnings catalyst; the article provides no evidence that it is financially material to NVIDIA.