Moxin, a leading Chinese domestic GPU vendor valued at ~300 billion RMB, breaks through with Hong Kong IPO listing

Previously, Moxie signed a strategic cooperation agreement with UB Tech to jointly establish Xixi Chuangzhi Technology Co., Ltd., focusing on R&D and commercialization of embodied AI chips. This marks another chip manufacturer entering the embodied AI track, following NVIDIA's collaboration with Unitree Robotics on humanoid robot development.

Recently, Moxie announced plans to issue overseas-listed foreign-invested shares (H-shares) and apply for listing on the main board of Hong Kong Exchanges and Clearing Limited. The company completed its listing on the Shanghai Stock Exchange's STAR Board on December 7 last year.

Given current market conditions and the company's operational status, Moxie's financing needs are urgent. To date, it has not achieved profitability, with cumulative losses exceeding 3.8 billion yuan from 2022 to 2025. Sustained high R&D investments require the company to maintain sufficient cash reserves; a successful Hong Kong listing would materially ease capital pressure.

From secondary market performance, Moxie has shown lackluster momentum since 2026 began. As of the latest close, its year-to-date gain stands at 32%, with a market capitalization of 308.2 billion yuan—significantly trailing the semiconductor sector's 76% gain in the same period.

It is evident that for Moxie to command a higher valuation post-Hong Kong listing, it must not only tap the embodied AI track to craft a new growth narrative, but also deliver more compelling core growth drivers.

Public records show Moxie's founders—Chen Weiliang, Peng Li, and Yang Jian—were all core technical leaders at AMD. Combined, they possess over 40 years of R&D experience in the GPU field. Chen Weiliang previously served as AMD's global head of GPU design, overseeing tapeout and mass production of 15 high-performance GPU chips. Peng Li was AMD's first Chinese female scientist. Yang Jian brings both international cutting-edge chip technology expertise and domestic commercialization experience.

This powerhouse founding team earned strong capital backing even at the early stage with only technical blueprints and team talent: within one month of founding, it secured angel financing close to 100 million yuan; within four months, it completed a pre-A round worth hundreds of millions led by Sequoia China. Over five years, Moxie completed seven funding rounds, raising a total of 12.64 billion yuan, backed by over 100 renowned institutions including state investment funds, Shanghai's innovation fund, and Sequoia China. The company boasts substantial capital reserves.

On December 7, 2025, Moxie—wearing the mantle of "China's second GPU leader"—listed on the STAR Board, with market cap briefly surpassing 330 billion yuan, becoming a market darling.

Behind the glittering surface lie visible signs of anxiety. Based on industry observation, the company currently faces dual predicaments: lagging technological progress and intensifying competition.

Technologically, NVIDIA has leveraged over 80% of the global GPU market share and tens of thousands of high-barrier patents to erect a near-monopolistic moat. Moxie's core product line directly benchmarks against NVIDIA's mainstream offerings: the Qiyun C series targets the A100/H100, while the Qisi N series targets the T4.

However, product benchmarking and technological parity are vastly different; surpassing NVIDIA poses immense challenges. Take the Qiyun C500, Moxie's flagship targeting the A100: it delivers FP16 peak compute of 240 TFLOPS with 64GB HBM2e memory. NVIDIA released the H100 in 2022—its FP16 compute reaches 1,979 TFLOPS, roughly 8 times the C500. In 2024, NVIDIA again unveiled the B200, doubling performance. Though Moxie's planned Qiyun C700 targets the H100 directly, NVIDIA maintains yearly generation cycles, leaving Moxie perpetually chasing the previous generation.

Process technology gaps represent insurmountable constraints. Moxie's C500 is produced by SMIC using 12nm; NVIDIA's H100 leverages TSMC's 4nm. Under semiconductor Moore's Law, process node gaps cannot be bridged short-term. Though Moxie's Qiyun C600—built entirely on domestic processes—successfully completed risk production in late 2025, SMIC's most advanced nodes remain several generations behind TSMC and Samsung. This leaves Moxie's chips with inherent disadvantages in transistor density and power efficiency at equivalent die size and power—hardware-level gaps that design optimization alone cannot overcome.

Competitively, the domestic-replacement wave has spawned numerous GPU startups, intensifying consolidation. Within just over a month from December 2025 to January 2026, four domestic GPU firms—Moore Threads, Moxie, Bitcomm, and Tianshu Intelligent—went public in rapid succession. Combined with already-listed players like Cambricon and Hygon, plus others like Enflame and Hanbor undergoing IPO preparation, over ten leading domestic GPU players now compete. The industry has fully transitioned from a supply-constrained blue ocean to a commoditized red ocean.

The GPU industry is unforgiving; not all participants survive. The sector demands extreme R&D investment and lengthy customer qualification cycles. Once leaders achieve scale with major customers, they establish powerful first-mover advantages and network effects. Currently, Moxie's customers concentrate on telecom operators and government AI compute centers; major internet customers remain in development. With purchasing decisions becoming cautious, the window for securing core large customers is rapidly narrowing.

In summary, despite holding "China's second GPU leader" status and a 308.2 billion yuan market cap, Moxie's competitive position is precarious amid difficult iteration and white-hot competition. Missteps could invite market elimination. This urgency explains why, merely half a year after A-share listing, it is rapidly orchestrating a Hong Kong IPO.

2025 was Moxie's first complete fiscal year on public markets, delivering impressive results: full-year revenue reached 1.644 billion yuan, surging 121.26% year-over-year; net loss attributable to parent stood at 789 million yuan, narrowing 43.97% from 2024's 1.409 billion yuan; operating cash flow improved 888 million yuan year-over-year; annual gross margin reached 56.51%, expanding 302 basis points as profitability quality continues improving.

Entering Q1 2026, growth persisted: the quarter achieved 561.9 million yuan in revenue, up 75.37% year-over-year; net loss narrowed to 98.84 million yuan, a 57.49% improvement year-over-year.

Beneath strong revenue growth and shrinking losses lies harsh reality: sustained cash flow pressure and surging capital needs.

Data reveal Moxie's 2025 operating cash flow was -1.26 billion yuan. Though significantly improved from 2024's -2.148 billion yuan deficit, the company's internal cash generation has yet to achieve positive circulation.

As of end-2025, Moxie had deployed over a dozen AI compute clusters, spanning national AI platforms, telecom operator platforms, and commercial centers across Beijing, Shanghai, Hangzhou, Changsha, and Hong Kong. The company's "1+6+X" strategy has achieved commercialization across six sectors—finance, healthcare, energy, education/research, transportation, and entertainment—while such large-scale infrastructure and industry expansion requires sustained heavy capital investment.

Meanwhile, Moxie maintains intense R&D spending. 2025 R&D expenses reached 1.027 billion yuan, up 14.04% year-over-year, representing 62.49% of revenue. Since 2020 founding, cumulative R&D investment has exceeded 3.2 billion yuan.

As noted, Moxie benchmarks its entire product line against NVIDIA, which now commands generational technology leadership. To narrow gaps and maintain pace with iteration cycles, Moxie must ramp R&D spending further. High R&D expenditure has become a relentless "cash-consuming machine."

Beyond easing financial pressure and securing R&D funding, the Hong Kong push is critical to ensuring continuous product iteration and long-term competitive durability. Currently, Moxie maintains a clear roadmap: the next-generation unified training-and-inference Qiyun C700 is in development; the Qixi X series targeting scientific intelligence launched formally in January 2026; embodied AI edge chips are planned to complete tapeout by end-2027 and reach mass production by 2028. Its joint venture with UB Tech, Xixi Chuangzhi, focuses on this frontier.

According to Moxie's disclosed Hong Kong IPO use of proceeds, H-share capital will primarily fund next-generation GPU R&D and commercialization, MXMACA software ecosystem development, industry-chain investment and M&A, sales and marketing upgrades, and working capital.

Hard-tech growth cycles are lengthy and unique. Industry patterns show high-end chips require 2-3 years from architecture design to tapeout verification; post-design production ramp-up alone needs over a year; final large-scale commercialization requires customer qualification, ecosystem adaptation, and channel development. Hong Kong IPO proceeds will provide dual funding safeguards for product R&D and technological iteration.

In summary, Moxie's rush to a Hong Kong IPO within half a year is far more than capital expansion—it is both a pragmatic response to short-term funding pressure and a core strategic pillar supporting long-term technological breakthroughs and global expansion.

Moxie's urgent pivot from A-share listing to Hong Kong within half a year reflects existential anxiety and positioning battles within China's leading GPU sector amid industry consolidation.

A generational technology divide that defies easy crossing, increasingly fierce sector consolidation, and unresolved losses compound, leaving Moxie in a perpetually taut competitive state.

A Hong Kong IPO will undoubtedly inject critical momentum for R&D iteration, business expansion, and ecosystem refinement, creating a safety net for development. Yet capital infusion is no panacea. The true core test for Moxie is whether it can leverage incremental capital to continuously refine technology, expand markets, and build unique competitive moats.