Nvidia’s Jensen Huang telling the Financial Times that “China is going to win the AI race” is not a provocation. It is a reading of deployment reality. China’s developer base, energy economics, and policy velocity are converging into a durable advantage in AI and electrification. Export controls will shape which chips train which models, but they do not stop the market from scaling. They redirect it. Here is where the capital is already compounding.
Huang’s point is simple: if you cut off access to a market that contains a huge share of the world’s AI developers, you drive that market to build a parallel stack. He called China “nanoseconds behind America in AI” and urged Washington not to “lose half of the world’s AI developers.” The political soundbite that Blackwell should be reserved for U.S. customers may play well on stage, but it misreads how fast China repurposes off-the-shelf compute, designs efficient models, and industrializes deployment. Every cycle of export restriction has accelerated domestic substitution in chips, cloud, and toolchains. Investors should assume follow-through, not friction.
The competitive difference is not only engineers. It is electricity, land, and logistics. China is building AI-ready power and data center capacity at a pace no other market can match, anchored by low-cost hydro in the southwest, a world-leading solar and wind buildout, and a vast grid-storage push. Cheaper, predictable power drives down the total cost of model training and inference. Add to that a gigantic base of mobile users and manufacturing depth, and you get an end-to-end AI economy that is hard to replicate. That is why Huang’s comments about energy costs and regulatory drag in the West resonated. This is not about policy slogans. It is about unit economics and speed.
1) Baidu (BIDU): ERNIE’s rapid iteration and Apollo’s robotaxi stack give Baidu the deepest full-stack AI in China; Reuters reports the company plans to take robotaxis beyond mainland China after winning driverless ride permits in major cities, a milestone that signals exportable autonomy. 2) Tencent (0700.HK, TCEHY): Hunyuan powers content, advertising, and enterprise tools across WeChat’s 1-billion-plus user ecosystem; with unmatched distribution and data, analysts see Tencent as the key monetizer of applied AI in China’s consumer internet. 3) Alibaba (9988.HK, BABA): Tongyi models are embedded across DingTalk and Alibaba Cloud, and the company’s domestic cloud remains the go-to AI infrastructure for many enterprises; the global impact is Alibaba Cloud’s expanding Southeast Asia footprint, where it is the leading non-U.S. hyperscaler. 4) PDD Holdings (PDD): Temu’s AI-driven merchandising and fulfillment are resetting global e-commerce price points across North America and Europe; sustained share gains show China-trained recommender systems exporting at scale. 5) Xiaomi (1810.HK): HyperOS enables on-device AI across phones, wearables, and home IoT, while the company’s move into EVs and AI glasses extends edge AI into mobility; the milestone is shipping a unified software layer that ties hundreds of millions of devices to one inference fabric. 6) BYD (1211.HK, BYDDY): With blade batteries and vertical integration, BYD surpassed Tesla in quarterly BEV sales and is now exporting to dozens of markets from Southeast Asia to Latin America; its global footprint is turning China’s battery leadership into EV market share. 7) NIO (NIO): A battery-swapping network of over 2,000 stations and continuous-range innovations make NIO the reference case for energy-as-a-service; pilots of 150 kWh packs and European swap deployments signal a differentiated model that travels. 8) CATL (300750.SZ): With roughly one-third global EV battery share, CATL is mass-producing fast-charging LFP and pushing sodium-ion into commercial use; its grid storage pipelines lower AI data center power costs and smooth renewable integration, a catalytic global impact. 9) SMIC (0981.HK): China’s top foundry is scaling mature nodes while enabling 7 nm-class domestic SoCs for leading smartphones; every incremental capacity ramp reduces reliance on foreign fabs and anchors a local AI silicon roadmap. 10) SenseTime (0020.HK): The SenseNova generative suite is already in market across finance, retail, and city applications; despite sanctions, model upgrades and customer wins show resilient demand for homegrown AI.
Nvidia has not applied for export licenses for its most advanced chips into China and says Beijing’s stance has shut it out. Even if Blackwell stays in the U.S., the direction is clear: Chinese hyperscalers will lean on a mix of domestically designed accelerators, optimized software, and model architectures that trade parameter bloat for efficiency. Huawei’s Ascend ecosystem, Baidu’s Kunlun, Cambricon designs, and a wave of inference-focused startups are not theoretical. They are entering procurement cycles. The playbook is the same one China ran in solar modules, 5G gear, and EV batteries: compress the cost curve, scale manufacturing, and integrate vertically. Investors should expect more domestic GPU alternatives, faster adoption of sparsity and quantization techniques, and a lot of training moving to lower-cost regions aligned with renewable capacity.
AI is already an energy story. China is adding renewable capacity at record pace, with provinces like Sichuan and Yunnan offering abundant hydro that pairs with AI compute parks. Storage is the bridge, and Chinese firms dominate the battery supply chain feeding both EVs and data centers. As battery costs fall and cycle lives improve, colocated storage will make Chinese data centers more resilient to peak pricing and grid constraints. Add cheaper construction and faster permitting, and the all-in cost per trained token drops. This is why global funds that once treated AI and energy as separate silos are now underwriting them together in China exposures. The capex flywheel is hard to stop once it starts spinning.
This wave is not confined within borders. BYD is building assembly footprints in Southeast Asia and considering capacity in Mexico to serve the Americas. PDD’s Temu has shown that Chinese supply chains plus AI curation can break into developed markets at speed. Alibaba Cloud is deepening its Southeast Asian presence, helping local developers train and deploy models closer to users. Reuters notes Baidu is preparing to take robotaxis overseas, signaling a playbook to export autonomy services. The theme is consistent: China’s AI and EV leaders are not just catching up. They are setting price and performance benchmarks in emerging markets, anchoring long-term customer relationships and standards.
Huang made it plain: he wants America to win by winning the world’s developers, China included. Policy may prevent that alignment, but demand inside China will not wait. Compute will get built, models will evolve, and applications will proliferate across industry. For global portfolios, that argues for exposure to China’s AI stack in equities that monetize deployment, not just training chips. It also argues for owning the energy and materials that make the stack cheap and reliable. If you are underweight because you think export policy locks China out, reconsider. The constraint is pushing scale into the domestic ecosystem with surprising speed.
This is a policy-heavy space. Export controls, entity lists, data localization rules, and antitrust probes can move prices. EV price wars and model commoditization will test margins. ADR structures and audit regimes are evergreen concerns. But the buffers are significant: world-class engineering, a giant home market, and a policy agenda focused on compute, power, and infrastructure. Companies doing the hard engineering work—batteries, chips, model tooling, robotics, logistics—are building moats in scale and cost that are hard to assail. The cadence of milestones in the list above is evidence. Each speaks to execution under constraint.
Huang’s remarks are a signal, not a surprise. China’s edge is not rhetoric. It is the compounding of developers, power, manufacturing, and policy. That flywheel is spinning faster, and it is pulling capital, talent, and customers into its orbit. If you want to understand where AI and electrification profits accrue next, look where the deployments are densest and the cost curves steepest. They point to China. Investors who position around that reality will find more upside than debate.
