Embrace NVIDIA, Confront Tesla: Can Geely (吉利汽车) Smart Driving 'Comeback' Succeed?

Financial turnaround masks a strategic gamble

Geely (吉利汽车) closed 2025 with headline numbers: annual sales topped 3.02 million units, revenue hit RMB 345.2 billion, and core net profit attributable to shareholders rose 36% year‑on‑year. Integration of Zeekr (极氪) after its U.S. delisting and deeper sharing with Lynk & Co (领克) cut administrative costs and unlocked margin levers. At the same time Geely chose to expense a larger share of R&D in the near term — reportedly Q4 R&D reached RMB 5.9 billion with an expense ratio up to 43% — a financial “deep dive” that sacrifices short‑term gross margin for balance‑sheet health and future capability.

A public deadline to match Tesla FSD — bold or precarious?

At the results briefing Gui Shengyue (桂生悦) set a striking public target: Geely’s smart‑driving software will reach parity with Tesla’s Full Self‑Driving (FSD) level within 2026. It has been reported that many Geely owners have voiced complaints about current driver‑assist experiences, making the pledge as much about reputational repair as technical ambition. But FSD itself is a moving target — Tesla iterates constantly — and the definition of “catching up” is not Geely’s to set. Is parity a statement of engineering confidence or a high‑risk flag planted for investors and competitors alike?

NVIDIA tie‑up answers compute but not the user test

Geely’s most visible move is the strategic alliance with NVIDIA (英伟达) announced at GTC: work on L4 capabilities using the DRIVE Hyperion platform and a stated intention to jointly define the next‑generation intelligent vehicle stack. That addresses the immediate compute bottleneck, yet real competition has shifted from raw silicon to experience. Will OTA refinement, edge‑case handling and user trust keep pace? And in the background, geopolitical realities matter: U.S. export controls and broader tech rivalry between Washington and Beijing add supply‑chain uncertainty to any heavy reliance on U.S. semiconductor partners.

2026: three battles that will decide if this is confidence or illusion

Geely has set modest growth targets for 2026 — about 3.45 million units — but success depends on three fronts. First, execution: H7 and subsequent smart‑driving releases must prove robust in daily use, not just in demos. Second, R&D conversion: the current wave of expensed investment must translate into measurable experience and safety gains, or it will look like sunk cost. Third, premium mobility: brands such as Zeekr and Lynk & Co must keep winning higher ASP (average selling price) buyers to sustain margin expansion. In the end, the market will judge Geely not on press conferences but on every OTA, every takeover, and every car handed to a customer. Is this a calculated comeback — or a gamble with too short a deadline?