On November 2, 2026, high-speed interconnect chip maker Yitai Intelligent Technology Co., Ltd. announced it had closed a Series A round of more than 200 million yuan. The A3 tranche was led by Wuxi Venture Capital, with Jintou Zhiyuan and Hongshan Capital participating.

The company was specific about where the money goes: volume delivery of its edge-side switch chips, and continued build-out of its senior talent bench. The stated goal is to break into the top tier of domestic high-speed interconnect chip suppliers.

The "A3" label is worth a note. It usually means the Series A was closed in several tranches, with A3 as one of them, rather than another step up in valuation. That staged structure is common among chip companies that need to spend in step with a production ramp.

Switch Chips: The Data Hub Inside the Car

To understand what this company does, start with the switch chip category itself.

A switch chip takes multiple streams of data and forwards each one to the port it belongs to — the intersection of a network. Inside a car, cameras, radar, the cockpit and the autonomous-driving domain controller are constantly passing data among themselves, and an automotive Ethernet switch chip decides who sends to whom, over which path, and at what priority.

Why Ethernet in a car? Traditional automotive buses are simple in both bandwidth and topology, which is fine for control signals for windows and lights. But once intelligent driving arrives, the data volumes from cameras and radar, and the east-west traffic between domain controllers, jump by an order of magnitude, and the old bus architecture no longer holds up. Automotive Ethernet was adapted for exactly this scenario.

The switch chip sits in a delicate spot in that chain: it does not determine computing power directly, but it determines whether data arrives on time. One faulty chip affects the entire data path. That is also why the barrier in this category is not the design itself but automotive-grade qualification and consistency at volume. The validation cycle before a chip gets into a car is long, and once it is in, it is hard to displace.

200 Million Yuan or 100,000 Vehicles: Which Number Matters More

The round is worth more than 200 million yuan, but the figure the company chose to lead with is a different one: 100,000 vehicles.

Wang Xingchen, vice president of Yitai Intelligent, said the company's first-generation automotive-grade switch chip, the ETLX610, is already in volume production across multiple vehicle platforms at a leading automaker, with more than 100,000 vehicles carrying it on the road, while other automakers are moving into production in stages. By his account, Yitai Intelligent is now the largest supplier of automotive Ethernet switch chips in China by shipments.

The weight of that number rests on two qualifiers: "multiple vehicle platforms" and "on the road." Winning a design slot is only the start. Getting into several platforms means the same chip has to run on different electrical and electronic architectures. And 100,000 vehicles on the road means it has cleared the lab and prototype stages and entered long-tail validation in real traffic. For an automotive-grade chip, that step is far harder than raising money.

Wang also pointed to where growth comes next: as commercial vehicles, driverless vehicles and eVTOL aircraft move into production, he expects sales of the company's automotive-grade chips to grow exponentially. eVTOL stands for electric vertical take-off and landing aircraft, and its weight and wiring-harness constraints are tighter than a car's — which happens to be where the next technology on the list comes in.

One Leg in Cars, One in National Projects

Alongside the automotive line, Yitai Intelligent runs an industrial-grade product line. Its first-generation industrial chip, the ETLX810, has been deployed in a major national project and is in volume use with customers in the power grid, high-speed rail and aviation sectors.

The company also noted two things: it has been awarded national projects, and it led the drafting of a national standard. The first is order book; the second is influence. In a category where overseas vendors have long defined the interfaces and standards, a seat at the standards table says the product is already embedded in the underlying infrastructure of core industries.

Running automotive and industrial lines in parallel spreads the risk. Automotive qualification takes years and ramps slowly, but a design win means years of supply. Individual industrial projects may be smaller, but they land faster. Two legs are what carry a chip company through the most cash-hungry years after a Series A.

Next Stop: Robots and Superpods

The company points to two directions ahead, both stated in the original report.

The first is humanoid robots. Infineon and Texas Instruments have already begun using single-pair Ethernet to drive robots. Single-pair Ethernet (SPE) carries both data and power over one twisted pair, which makes harnesses lighter, thinner and cheaper. Robots have many joints and little room for routing, so the fit is natural. As the original report puts it, in future edge-side scenarios, general-purpose Ethernet offering high bandwidth, low latency, high reliability and determinism is expected to become the optimal interconnect solution.

The second is superpod interconnect in AI networks. A superpod links hundreds or thousands of accelerator cards over high-speed interconnect into what behaves as one large logical machine, coordinating them as if they sat inside a single chip. How traffic is forwarded and scheduled between cards again comes down to switch chips. With the new funding, the company plans to move aggressively into this area.

The gap between the two is not small. Automotive Ethernet demands determinism and reliability; superpod interconnect demands extreme bandwidth and minimal latency, and the engineering behind each is not fully interchangeable. But the underlying forwarding architecture, SerDes and switch scheduling can be reused, which is where the "high-speed interconnect chip" label comes from. What is being sold is not one specific chip but a capability to move data around.

On the team side, the company said it has recruited a co-founder-level head of automotive commercialization, a top international architect and a high-speed interconnect team in recent years. Founder Zhong Jianfeng framed it this way: "Building a startup is a process of accumulating talent. You have to keep raising your strengths as you grow, and keep bringing in the key people who have a global perspective and want to change the world."