A sharp uptick in Chinese AI releases and a high-profile hardware decision by Waymo are converging to accelerate the commercial phase of artificial intelligence and autonomous mobility. In recent weeks domestic firms rolled out generational advances across video generation, large models and compute infrastructure, while Waymo’s sixth‑generation Ojai autonomous taxi began trials in San Francisco and Los Angeles on a lower‑cost, China‑made chassis engineered for harsher weather.
The Chinese technology wave is broad. ByteDance’s Seedance 2.0 has reset expectations for AI video generation, while open releases such as Zhipu’s GLM‑5 are running large inference clusters on domestically produced chips — evidence that China’s push to “de‑NVIDIA‑ify” the stack is making practical progress. Leading vendors including DeepSeek, Kimi, Alibaba Tongyi and Baidu Wenxin shifted their recent upgrades away from raw parameter count toward industry adaptation and longer context handling; Baidu’s Wenxin 5.0 now lists about 2.4 trillion parameters and DeepSeek has extended context windows toward one million tokens.
That combination of model refinement and indigenous compute matters because 2026 is shaping up as the year AI moves from experimental prowess to routinised commercial deployments. Chinese policy now explicitly targets sustained growth in AI compute infrastructure — the New Generation AI Development Plan for 2026–2030 calls for annual investment in compute to rise at least 20% — while market forecasts put the global AI server market well into the hundreds of billions of dollars. Domestic advances in chips, optics and servers are therefore not just national pride; they are the backbone of scaleable, lower‑cost solutions that feed both enterprise AI and emerging autonomous mobility services.
Waymo’s adoption of a China‑manufactured chassis for its Ojai taxi is a practical illustration of those shifts. The company’s engineering leadership frames the new platform as the core engine for the next stage of expansion, aiming to open public services within the year and to chase a projected $25 billion autonomous ride‑hailing market by 2030. The choice signals that Chinese hardware suppliers have reached sufficient maturity to meet high reliability, cost and weather‑resilience thresholds required by global autonomy fleets.
For investors tracking the industrial effects of this transition, index‑based vehicles are already positioning to capture the supply‑side winners. The Tianhong CSI Artificial Intelligence Thematic Index Fund (A class: 011839, C class: 011840) replicates the CSI AI thematic index and holds a concentrated mix of components across the AI value chain. As of 11 February 2026 the index’s top ten weights — including optics players, AI chipmakers, server and algorithm specialists — account for 55.72% of the index, with sector exposures that read like a map of China’s AI supply chain: optical modules and communications (21.14%), AI chip/processors (14.51%), servers/compute (4.70%), algorithms/NLP (4.36%), computer vision (4.11%) and application/IP platforms (5.22%).
The fund’s performance has been strong in the most recent market cycle: since inception it suffered steep losses in 2022 but rebounded with double‑digit returns in 2023 and 2024 and a very strong 2025 (+64.98% A class, +64.66% C class). Assets under management stood at 33.94 billion yuan at the end of 2025, with management and custody fees totalling 0.6% and a C‑class sales service fee of 0.25%. Tracking errors for the past year were modest (A class 1.34%, C class 1.02%), indicating relatively tight index replication for an off‑exchange AI thematic ETF.
The industrial and financial narratives reinforce one another. As models migrate toward verticalised, industry‑specific deployment and as domestic compute costs fall, demand for optical components, chips, servers and software solutions rises. That creates a reinforcing cycle: applications drive demand for compute and data, compute scale reduces costs and broadens addressable markets, and resilient domestic supply chains reduce geopolitical and logistical fragility. Yet commercialisation will not be frictionless: regulatory alignment, safety verification for autonomous fleets and competitive pressure on margins are immediate constraints.
Market participants should therefore expect a bifurcated near term. Companies that can translate model capabilities into robust, certified enterprise or mobility services will capture disproportionate revenue growth; component suppliers that secure long‑term contracts and scale manufacturing will benefit from improving unit economics. Passive funds that mirror an AI index offer accessible exposure to this structural shift, but investors must weigh concentrated sector risk, tracking error and the fund’s fee structure against a backdrop of rapid technological and policy change.
This moment is less about a single technological leap than about systems coming together — algorithms, chips, hardware suppliers, regulatory frameworks and financing — in ways that make large‑scale, real‑world AI and autonomous services commercially viable. For global audiences, the twin signals from China’s AI releases and Waymo’s supply‑chain choices indicate that cost, resilience and vertical application are now front and centre in the race to industrialise artificial intelligence.