China's tech export machine: cheap EVs, open AI models, one chokepoint at a time

In June, a purpose-built LNG-powered carrier called the BYD Zhengzhou docked at the Port of Melbourne with 5000 newly built electric vehicles on board, the largest single EV shipment ever sent to Australia. Ian Verrender, chief business correspondent at ABC News, wrote in RNZ that the cargo illustrated a changing of the guard in the automobile industry, in Australia and globally.

Australia works as a test case. It has had no domestic car industry since its own collapsed a decade ago, and it has few trade barriers. That makes it a useful gauge of consumer taste. The numbers there are brutal for the old guard. In the first half of 2026, Mazda shed 17 percent of sales compared with the same period a year earlier. Toyota was down 21 percent. Subaru and Mitsubishi dropped 25 percent each, and Nissan was off 32 percent.

Chinese suppliers moved the other way. BYD sales jumped 124 percent, Chery leapt 77 percent and Geely soared 495 percent. All three started from a much smaller base, which exaggerates the percentage gain. The trend itself is not in dispute.

The shift is not only about price. Riz Akhtar, founder of the research group carloop, told ABC News he drives one of the cars. "It costs around $40,000. It's quiet, it's comfortable, drives well and is loaded with technology," he said. An equivalent Japanese- or European-built vehicle, he added, would cost more than 25 percent more. "Consumers aren't stupid. They're not going to spend money they don't need to."

Batteries, minerals and a 90 percent share

Underneath the vehicle sales sits a supply chain China has spent decades assembling. The same RNZ analysis notes that China dominates the minerals essential to battery production, among them lithium, cobalt, manganese and graphite, and turns out almost all the world's anode and cathode material. It now accounts for 90 percent of global battery production.

That dominance is the backdrop to a fight now playing out in Washington. MIT Technology Review reported in September that the US is setting records for growth in its energy storage market, largely on the back of cheap Chinese batteries. In late August, the Trump administration declared a national emergency that effectively bans Chinese batteries from grid-scale storage systems. Import taxes on batteries rose to 25 percent in January, up from 7.5 percent.

Analysts quoted by MIT Technology Review expect the ban to slow deployment of grid-connected storage in the near term. Shan Tomouk of Benchmark Mineral Intelligence called the outright ban "a bit of a surprise" that "does create a bit of concern for domestic players in the US." Isshu Kikuma of BloombergNEF said some projects may need alternative cell sources that cost more than Chinese imports. "Worst case, those projects could get canceled," he said.

The US could have enough domestic capacity by about 2030, though some factories may not ramp up fully. That would push real self-sufficiency into the late 2030s.

Open models and a White House at war with itself

The second front is software. Chinese AI models such as Moonshot's Kimi K3, Qwen, DeepSeek and GLM are generally open-weight, meaning the model software and final parameters are released, but not the training data and methodology. MIT Technology Review reported in July that Kimi appeared to rival models from OpenAI and Anthropic, which are not free, and that the release split President Donald Trump's AI advisers into open factions.

David Sacks, Trump's AI and crypto czar until March, branded Anthropic's models "lobotomized" and "woke," and argued that Chinese models are popular because they come with fewer restrictions on use. Emil Michael, a top Pentagon official, called OpenAI's new head of strategic futures a "supreme village idiot." Dean Ball, a former Trump AI adviser now at OpenAI, criticised a new White House review process as a "de facto licensing regime for frontier AI."

The Guardian reported on 1 August that Microsoft, Nvidia, Palantir and Meta signed a letter urging lawmakers not to restrict open models, while Nvidia CEO Jensen Huang went to Capitol Hill to lobby on the same side. Anthropic and OpenAI argue the opposite, saying Chinese-made models pose security risks. Treasury secretary Scott Bessent floated sanctions over alleged intellectual property theft. Commerce secretary Howard Lutnick received letters from startup founders asking him not to cut off access.

The same Guardian piece notes the FCC banned humanoid robots from China over alleged national security risk, and that a report of Chinese mass production of specialty AI chips erased $1tn in market value from other chip makers.

Trust is the open question

Writing in The Conversation, a China expert and open-source researcher argues that Chinese open-weight models are built to align with government censorship and propaganda, and that some licences carry extra obligations. The Kimi K3 licence, for example, requires large licensees to "prominently" display the model name. MiniMax M3's community licence lists uses prohibited by "applicable laws or regulations."

Bilingual licences developed under Chinese government guidance, including the Mulan Permissive Software License version 2.0 and the OpenAtom Model License version 1.0, are written in English and Chinese. The Chinese version prevails when the two conflict, especially in Chinese jurisdictions.

Elsewhere, the legal pressure runs the other way. A US prosecutor told a Brooklyn federal court in September that Huawei was a criminal enterprise that stole trade secrets from five US companies, including router source code from Cisco and a robotic arm from T-Mobile. Defence lawyer Brian Heberlig told jurors: "It's about competition, not conspiracy. Innovation, not theft." A Chinese foreign ministry spokesperson said Beijing "firmly opposes the US side's suppression and containment of Chinese enterprises."