Wang Dongsheng spent decades building BOE into one of the world's largest display makers. Six years after stepping down from that chair, the company he helped start is asking Hong Kong investors to back a different, screenless bet. The pitch is RISC-V — the open instruction set that China increasingly treats as a route around a closed chip architecture.
Key takeaways
- Beijing ESWIN Computing Technology filed a Hong Kong IPO prospectus on May 30, 2025, with CITIC Securities and CITIC Construction Investment International as joint sponsors.
- By 2024 revenue, ESWIN is China's largest RISC-V full-custom solution provider and its fourth-largest RISC-V main-control solution provider, with more than 100 commercialized system-level solutions and over 100 global customers.
- Revenue ran at about 2.0 billion yuan (2022), 1.75 billion (2023), and 2.03 billion (2024) — roughly US$281 million / HK$2.2 billion in 2024 — while losses totaled 1.57 billion, 1.84 billion, and 1.55 billion yuan, a three-year hole near 4.95 billion yuan (≈ US$690 million / HK$5.4 billion).
- A single customer accounted for 76.8%–82.1% of revenue across 2022–2024, a concentration risk the prospectus flags openly.
- Cumulative fundraising exceeded 9 billion yuan (≈ US$1.25 billion / HK$9.8 billion), with backers including the National Integrated Circuit Fund Phase II and IDG Capital.
What ESWIN actually does
ESWIN is not a chip maker in the classic sense of shipping standard processors. It designs system-level solutions built on self-developed RISC-V cores, pairing proprietary IP modules with software to serve two markets: smart terminals (consumer electronics, IoT) and embodied intelligence (smart vehicles, robots, industrial clients). The company says it holds 500-plus IP modules, 20-plus serialized RISC-V cores, and has applied for 1,500-plus patents.
That positioning matters because RISC-V is an open standard, free of the licensing structures that govern x86 and Arm. For a Chinese compute company, that openness is less a technical curiosity and more a structural hedge: it removes a foreign licensor from the critical path. ESWIN's bet is that the same architecture can scale from a smart-display controller to the compute brains of a robot or a car.
The money picture
The prospectus is unusually candid about the cost of that bet. Revenue was roughly flat across the three years — about 2.0 billion yuan in 2022, 1.75 billion in 2023, and 2.03 billion (≈ US$281 million / HK$2.2 billion) in 2024. Profitability is further away: the company posted losses of 1.57 billion, 1.84 billion, and 1.55 billion yuan, stacking up to a cumulative deficit near 4.95 billion yuan (≈ US$690 million / HK$5.4 billion). Gross margin also swung sharply, from 25.9% in 2022 to 15.4% in 2023 before recovering to 17.7% in 2024.
None of this is unusual for a pre-profit semiconductor design house chasing a new architecture. What is unusual is the transparency: the document lays out the burn rate in plain terms rather than burying it. Investors underwriting the "RISC-V first" story are being told, up front, that the path to profit is unproven.
Why RISC-V is the strategic story
Strip away the financials and the filing is a referendum on a national direction. China has spent years hedging against restrictions on advanced chips and the architectures that run them. RISC-V offers a license-free foundation that domestic firms can extend without asking permission, and ESWIN is among the most visible attempts to turn that freedom into shipped products at scale.
The market context the prospectus cites is steep growth: Frost & Sullivan, commissioned for the filing, estimates the global RISC-V main-control solution market grew from 1.1 billion yuan in 2020 to 56.5 billion yuan in 2024 and could reach 440.4 billion yuan by 2029. We treat those figures as optimistic, vendor-sponsored projections rather than settled fact. The directional point — RISC-V moving from microcontrollers into higher-value compute — is what makes ESWIN's listing attempt a watch item.
The risks the prospectus doesn't hide
Beyond the losses, the document surfaces a concentration problem that would worry any analyst. A single customer supplied between 76.8% and 82.1% of revenue in each of the three years. That level of dependence means one contract renewal, or one demand swing, moves the whole company. The prospectus also notes ESWIN remains unprofitable and that its RISC-V roadmap depends on ecosystem and software maturity that is still maturing industry-wide.
There is also the question of timing. The filing arrives in a market that has grown skeptical of cash-hungry hardware names, and alongside a sister listing: ESWIN Materials, the group's wafer business, separately pursued a STAR Market listing on the Shanghai exchange. The two tracks show a group trying to finance both the silicon and the architecture at once.
According to Li An, chief scientist at BrainNet (脑机网), ESWIN's filing matters less as a financial event than as a signal that China is willing to take RISC-V from embedded devices into the public markets as a compute architecture. Li An's comments reflect his own views and do not represent NeuroAI's editorial position or an endorsement by BrainNet (脑机网).
Honest limitations
This article draws on ESWIN's IPO prospectus and filings reported by financial outlets; it is a pre-listing snapshot, not an audited public track record. We have not modeled the likelihood of HKEX approval or post-listing performance, and we do not know whether the listing has since completed. The RISC-V market forecasts quoted are from Frost & Sullivan and were commissioned for the filing, so they should be read as company-sponsored, not independent. The identity of the dominant customer is not disclosed in the materials we reviewed, so we cannot assess that dependency's true stability.
What readers can do now
- Watch whether the HKEX listing clears and at what valuation — that price will signal how much public capital believes China's RISC-V compute story is worth.
- Track RISC-V adoption beyond low-end embedded chips into AI-edge and robotics, where ESWIN and peers are actually shipping, rather than treating the architecture as a distant rival to x86 or Arm.
- For engineers and product teams, evaluate the RISC-V software and toolchain maturity directly before committing a roadmap to it; the ecosystem, not the cores, is still the practical bottleneck.
