Two Chinese deep-technology startups announced large funding rounds on October 28: the Hangzhou AI inference-chip designer Nano-core Chip, which raised more than RMB 1 billion in a Series C2 round, and the brain-computer interface company NeuroXess, which raised RMB 700 million in a Series B. Together the rounds exceed RMB 1.7 billion. TechNode reported both deals, citing the Chinese outlets Zhidx and PEdaily, and other Chinese-language reports add investor names and technical detail.

Neither company disclosed a valuation, and neither has yet shown commercial sales. Both said the money will go toward getting a first product through the next stage: customer testing for the chips, and regulatory registration and factory construction for the implants.

Nano-core: three chips taped out

Nano-core Chip, which the Chinese press also calls Micro-Nano Core, was founded in 2021 and incubated at a Peking University research institute in Zhejiang province, TechNode reports. JianTou Investment led the Series C2, with Hangzhou Capital and Kaitai Capital also taking part. The Chinese outlet Wallstreetcn lists further participants, including Xiaochi Capital, Yantai Caijin and Guotou Taikang, and says several existing shareholders increased their stakes.

The round follows a RMB 1 billion Series C1 announced in September. Wallstreetcn reports that the company says it has raised more than RMB 2 billion in two months and RMB 3 billion in six months. The company plans to spend the new money on product development and on stocking its supply chain for volume production.

TechNode says the company’s first three chips have completed tape-out, the stage at which a finished design is sent for manufacturing. Wallstreetcn says the three belong to two product lines, called LP-CIM and PCIe-CIM. Customer sampling and initial volume shipments are planned for early 2027, according to both reports.

What the chips do and what is unproven

The chips are built around compute-in-memory, a design that performs calculations inside or next to memory to cut the energy and bandwidth cost of moving data between memory and a processor. Wallstreetcn describes Nano-core’s architecture, which it calls 3D-CIM, as combining stacked DRAM near-memory computing, SRAM compute-in-memory and RISC-V processor cores. The company targets AI inference, the stage at which a trained model answers queries, on devices such as AI phones and PCs and in cloud computing centers.

Wallstreetcn reports that, in earlier tape-out tests, the company says its SRAM compute-in-memory design delivers 4 to 6 times the compute density and 5 to 10 times the energy efficiency of conventional architectures. It adds that the company has not published full comparison specifications or test conditions. These are company claims.

The same report separates three stages that are easy to blur: simulator verification, solution testing, and physical sample testing. Nano-core has completed several rounds of simulator verification and says some projects are in deep joint debugging with phone platform makers, device manufacturers and embodied-AI companies. It has not named sample customers, announced volume purchase orders or disclosed revenue, Wallstreetcn says.

The report also places the company among rivals that are further along. Houmo Intelligence said in mid-October that its M30 and M50 chips are in mass production, that the M50 has more than 150 product design wins, and that its first 3D compute-in-memory chip, the A20, is in production. Zhicun Technology reports cumulative shipments above 10 million chips, and Yizhu Technology announced more than RMB 1 billion in financing on October 23, according to Wallstreetcn.

NeuroXess: a fully implanted system enters a registration trial

NeuroXess’s Series B was co-led by SDIC Juli, Shenzhen Capital Group and Qinzhi Capital, TechNode reports. The Chinese outlet IT Home, republished by Phoenix, and 36Kr list other investors including SMIC Capital, IDG Capital, Huafu Securities and Jiangxi Financial Holdings, and say the existing shareholders Shanda and Zhongping Capital increased their stakes.

TechNode says the money will go toward multicenter clinical trials, regulatory filings, manufacturing capacity and next-generation products. It describes the company’s implantable systems as pairing flexible electrodes with wireless power and data transmission, designed for decoding movement and speech and for controlling external devices. The company has completed 65 clinical cases, TechNode says, and launched a registration trial in July involving nearly 20 hospitals.

36Kr gives more detail in an interview with Tao Hu, whom it identifies as NeuroXess’s founder and chief scientist. The company calls its product a three-full system: fully implanted, fully wireless and full-function. According to 36Kr, all core components, including the battery, sit inside the body, with the electrodes resting on the surface of the cortex beneath the dura mater, not penetrating brain tissue, and the heat-generating parts placed under the skin of the chest.

36Kr reports that the July trial covers the system as a Class III medical device, is led by Huashan Hospital of Fudan University with Mao Ying as principal investigator, and targets severely ill patients with high-level spinal cord injury, aphasia or blindness. Those are the company’s plans and Tao’s statements, not trial results. No outcome data from the registration trial appear in the sources Lens opened.

TechNode adds that NeuroXess is building a manufacturing facility of 14,300 square meters with planned annual capacity of 10,000 systems. IT Home reports the company aims to bring products to scaled deployment in the medical market within two years.

Tao told 36Kr that dozens of clinical datasets are not enough because a model tuned to one patient may fail for the next, and that hundreds or thousands of cases are needed before algorithms transfer between patients. That caveat comes from the company’s own founder.

Who is paying

Tao told 36Kr that heavy state participation reflects the elevation of brain-computer interfaces to a national strategy, and 36Kr describes state capital as dominant in the round, alongside industrial investors from semiconductors, medical devices and materials. He also said that investors who three or four years ago saw the field as science fiction now focus on clinical results, patient benefit and domestic supply chains.

NeuroXess has raised before. Yicai reported in December 2022 that the company closed a round led by Zhongping Capital worth hundreds of millions of yuan, with earlier backers including Shanda Investment and Sequoia Capital China taking part. Zhongping and Shanda are among the shareholders reported to have added to their stakes in the new round.

Two smaller deals the same week

TechNode reported two further rounds in the same week, each from a single source. On October 30 it said Kai-Fu Lee’s 01.AI secured investment worth hundreds of millions of yuan from state-backed entities in Hangzhou and its Linping district, citing IT Home. Under the arrangement, 01.AI will base its headquarters in Linping and build a center to help companies adopt AI. The article gives no exact amount or investor names.

TechNode says 01.AI has moved away from consumer services toward enterprise AI, that its public model platform stopped offering trials and API services in September, and that Lee says the firm tailors Chinese open-weight models including DeepSeek, Qwen and GLM for business clients. Roughly half of revenue comes from outside China, according to the report. In July, Bloomberg Law carried a Bloomberg report that 01.AI, which that report describes as headquartered in Beijing, was raising money ahead of a planned Hong Kong IPO in 2027.

On October 29, TechNode reported, citing The Beijing News, that Jingshuo, a subsidiary of the autonomous-driving company WeRide that builds data infrastructure for physical AI, raised tens of millions of dollars in a pre-Series A round. Investors include Vision Knight Capital, Hong Kong Science and Technology Parks Corporation and Sirius Capital. Jingshuo’s CEO is Huo Da, WeRide’s first employee, and WeRide remains the majority owner, so Jingshuo’s results stay in WeRide’s consolidated accounts.

The funding backdrop

The rounds arrive while listed Chinese technology shares are under pressure. Reuters, in a report published October 28 by The Daily Star, says the STAR50 index of mainland AI and semiconductor companies has fallen 40 percent in less than four months and that tech equity fundraising is likely to slow. Hong Hao of Lotus Asset Management is quoted as saying the sell-off will halt many companies trying to list.

Lens analysis: private rounds from investors with state ties, as in the Nano-core, NeuroXess and 01.AI deals, may be a way for companies to raise money while the public market is weak. The sources Lens opened do not say this directly, and none of the companies discussed here has announced a listing date.

What to watch

For Nano-core, the test comes in early 2027: whether physical chips reach named customers, and whether purchase orders follow. For NeuroXess, the registration trial’s results and the regulator’s decision are the milestones, and no timetable for either appears in the sources Lens reviewed. For both, valuations and revenue remain undisclosed, so the size of the rounds says more about investor appetite than about commercial traction.

Sources

Image: 20260424 West Lake and Hangzhou Skyline.jpg by Windmemories, CC BY-SA 4.0, via Wikimedia Commons. Illustrative photo; it does not show the events described.