NextFin News - Two Chinese chip founders are now paper billionaires, and the company that made them rich has never turned a profit. Enflame Technology listed on Shanghai's STAR Market on September 11, 2026, at 142.18 yuan per share, valuing the eight-year-old AI chipmaker at about 61.2 billion yuan ($9.1 billion) despite cumulative losses of more than 4.3 billion yuan over the past three years and revenue of just 990 million yuan in 2025. Co-founders Zhao Lidong and Zhang Yalin, both AMD veterans, jointly control 28.14% of the company — a stake worth roughly 17.2 billion yuan on paper. The listing forces investors to answer a question that goes well beyond one company: has China built a capital market that prices national champions ahead of profits, and how long can that pricing hold?
The Deal: A $9 Billion Valuation Built on Losses
Enflame's listing notice, filed after market close on September 9, sets the terms plainly. The company sold 43.04 million new shares — 10% of its enlarged capital — raising about 6 billion yuan, with no existing shareholders cashing out. CITIC Securities acted as sponsor and joint lead underwriter, alongside Guotai Haitong Securities and GF Securities. The issue price implies a diluted 2025 price-to-sales ratio of 61.8 times, which the company notes is below the peer average of 93.5 times. That comparison is meant to reassure. It also concedes the point: the valuation rests entirely on sales, because there are no earnings to multiply.
The financial trajectory is steep in both directions. Revenue grew from 301 million yuan in 2023 to 722 million yuan in 2024 and 990 million yuan in 2025, a three-year compound annual growth rate of roughly 81%. In the first half of 2026 alone, revenue reached 1.12 billion yuan, exceeding the entire prior year. Losses, however, remain large: net losses attributable to the parent were 1.665 billion yuan, 1.51 billion yuan and 1.164 billion yuan across those three years, leaving an accumulated deficit of 4.441 billion yuan at the end of 2025. Cash has been burning for years; operating cash flow was negative in each of the three reporting periods, totaling nearly 4 billion yuan of outflow.
The spending is deliberate. Cumulative research and development outlay reached 3.676 billion yuan from 2023 to 2025 — R&D ran at 408% of revenue in 2023, 182% in 2024 and 115% in 2025. 643 engineers make up 76.7% of the workforce. Gross margin has improved, from 22.6% in 2023 to 31.8% in 2025, but it sits roughly half the peer average of about 58%. Before the public markets ever saw the company, the private capital engine had already done its work: Enflame raised about $746 million across six rounds through a September 2023 Series D, reaching a $2.5 billion private valuation by October 2021. The IPO is not the beginning of the funding story. It is the exit for it.
Chairman Zhao Lidong told investors on the IPO roadshow that, weighing orders in hand, delivery schedules and R&D budgets, Enflame expects to reach consolidated profitability in 2026 or 2027. He projected revenue of 2.3 billion to 3 billion yuan for the first three quarters of 2026, implying year-over-year growth of up to 455%. Even management's optimists do not claim the path is smooth. Huang Lichong, president of Huisheng State-Owned Capital, warned that narrowing losses should not be confused with a profitability inflection point: the test is whether orders pass acceptance and convert into recognized revenue, whether wafer foundry and packaging costs hold, and whether key-customer pricing turns growth into gross profit. In his reading, 2026 could approach breakeven only if delivery and margins clearly beat expectations; 2027 is the neutral verification window; and continued losses beyond 2027 remain a scenario that must be priced.
There is one more concentration risk embedded in those numbers. In 2025, sales to a single customer, Tencent, accounted for 83.79% of revenue. Tencent is also the largest external shareholder, holding 20.26% of Enflame alongside its affiliates. The company that is buying the chips also owns a fifth of the company and helped anchor the valuation. That is strategic alignment. It is also a fragility: if Tencent's procurement shifts, both revenue and the equity story move together.
The Capital Engine Behind the Valuations
Enflame is not an anomaly. It is the latest product of a capital architecture that China has assembled around semiconductor self-reliance. The STAR Market, launched in 2019 as Shanghai's answer to the Nasdaq, was built with listing standards tailored to pre-profit technology companies. In 2026 it has become the launchpad for a cohort of domestic chip champions whose market values are being set by strategic demand rather than discounted cash flows. The benchmark tells the story: a fund tracking the STAR 50 index gained 27.2% year-to-date through the end of July 2026, far outpacing the broader Chinese equity market.
The pattern is now familiar. CXMT, the Hefei-based DRAM maker, priced its IPO at 8.66 yuan and raised 57.92 billion yuan ($8.6 billion) — Asia's largest offering of 2026. On its first trading day, July 27, the shares closed at 49 yuan, a 466% jump that lifted the company's market value to roughly 3.3 trillion yuan, briefly making it China's most valuable listed company, ahead of Industrial and Commercial Bank of China's 2.6 trillion yuan. Moore Threads, the first of China's four domestic GPU challengers to list, surged about 468% at the open on its December 2025 STAR debut and saw its market value exceed 300 billion yuan. Biren Technology, which listed in Hong Kong on January 2, 2026, gained 75.82% on day one and crossed 100 billion Hong Kong dollars. MetaX, the fourth member of the group, listed on December 17, 2025. Most of the cohort was loss-making at the time of listing; CXMT is the exception, having swung to a first-quarter 2026 operating profit of 35.43 billion yuan from a loss of 2.83 billion yuan a year earlier.
The demand is not accidental. A network of state capital sits behind it. China's National Venture Capital Guidance Fund aims to mobilize around $144 billion for early-stage and strategically important technologies. Big Fund III, the latest phase of the National Integrated Circuit Industry Investment Fund, adds approximately $47.5 billion earmarked for the semiconductor ecosystem. State-owned investment groups deployed nearly $9 billion into the broader equity market one week before the CXMT offering, following a technology sell-off. Mutual funds have been directed to increase A-share holdings, insurers are allocating more new premium income to equities, and central-bank-backed liquidity facilities provide another layer of support. The Shanghai Stock Exchange accepted the STAR Market IPO filing of XPHOR, a silicon-photonics company planning to raise 2.43 billion yuan, and greenlit InnoGrit, a storage-controller developer — the pipeline is not thinning.
What this means is structural. China is not merely subsidizing chip companies; it is building an end-to-end capital engine — government funds, venture capital, a listing venue with permissive standards, and directed institutional demand — designed to fund, list and scale strategically important firms regardless of current profitability. In that system, a company's value is partly a function of its strategic role, and the market is being asked to price that role ahead of the earnings that would normally justify it.
The Second-Order Consequence: Wealth Before Earnings
The first-order effect of this system is obvious: chip companies get funded. The second-order effect is what Enflame's listing makes visible — founders and early backers can become billionaires years before their companies generate sustainable profits, because the exit is no longer a profitable business. It is a listing.
That changes the incentive structure across the sector. When the capital market rewards strategic positioning ahead of earnings, the rational move for founders is to optimize for listing readiness — revenue scale, patent counts, customer logos — rather than unit economics. Enflame's own numbers show the trade: revenue nearly tripled in three years while R&D consumed more than three times the cumulative revenue. The company is racing to scale before the window closes, and the window is defined by policy as much as by technology.
There is also a cross-market transmission that most investors are not pricing. The same capital engine that lifts Chinese chip valuations compresses the timeline for domestic substitution. Chinese cloud providers and internet giants, facing US export controls on advanced accelerators, are being pushed toward local suppliers. That demand is real and growing — but it is also policy-driven. If policy support or procurement mandates ease, the revenue that underpins these valuations could prove less durable than the capex cycle suggests.
These memory chip businesses are sustainable, but the great margins and net profitability we're seeing today are not sustainable and have to normalize over a cycle.
That warning came from Theodore Shou, chief executive of Yiyi Capital, speaking on the occasion of CXMT's July debut. It applies with equal force to the AI accelerator cohort: the current margin and valuation environment is a product of a supply squeeze and a policy tailwind, both of which are cyclical even if the strategic direction is not.
The Counter-Thesis: This Time, the Window Is Real
The strongest case against skepticism is straightforward, and it is backed by the market's own behavior. US export controls have closed off access to Nvidia's most advanced chips for Chinese buyers, creating a protected domestic market that did not exist a decade ago. Chinese data centers cannot wait for parity; they need chips now, and they will buy domestic. Enflame's backlog supports this: the company expects revenue of 2.3 billion to 3 billion yuan in just the first three quarters of 2026, more than triple its full-year 2025 revenue. Strategic placement in the IPO included Xiaomi, GigaDevice, Fujian Fuzhou F-Tech Microelectronics, ZTE and the National Council for Social Security Fund — industrial buyers and long-term capital, not just momentum traders. If domestic substitution is a multi-year structural shift rather than a sentiment cycle, today's valuations may look cheap in hindsight.
The counter-thesis has merit on the demand side, but it does not answer the supply side. Making chips is not the same as designing them. China's domestic GPU makers still lag global leaders in performance and, more importantly, in software ecosystems — Nvidia's CUDA carries nearly two decades of developer lock-in. Enflame's gross margin of 31.8% is roughly half the peer average, a sign that it is still paying to prove its products. And the customer concentration is acute: 83.79% of revenue from one buyer means the company's fate is tied to a single procurement relationship that is also a shareholder relationship.
The falsifying signal is specific. If Enflame fails to deliver consolidated profitability by the end of 2027 — the outer edge of management's own guidance — while gross margin remains below 40% and Tencent's share of revenue stays above 70%, the thesis that the capital engine can indefinitely defer the profitability test would be wrong. Profitability, not revenue scale, is the threshold the market is being asked to ignore. Once ignored long enough, it becomes the only metric that matters.
What Comes Next: Three Horizons
Short term (the first five trading days): Enflame faces no daily price limit during its first five sessions, with only 4.16% of post-IPO capital in unrestricted free float. That combination produced the violent first-day moves seen in CXMT and Moore Threads. Expect volatility, not price discovery. The online subscription winning rate of 0.0246% and retail oversubscription above 6,000 times signal a crowd chasing the pop, not the fundamentals.
Medium term (2026-2027): The profitability test arrives. Management has guided to break-even in 2026 and profitability no later than 2027. The base case is that Enflame narrows losses through 2026 and reaches consolidated profitability in 2027, supported by the revenue ramp and continued Tencent procurement. The upside case is that AI capex from Chinese cloud providers accelerates faster than expected, pulling profitability forward into late 2026 and validating the 61.8x sales multiple. The downside case is that margin expansion stalls, R&D spending stays above 100% of revenue, and the profitability timeline slips — at which point the valuation re-rates toward companies that have already proven earnings.
Long term (structural): The capital engine itself is the durable part of this story. Even if individual names re-rate, the architecture — guidance funds, Big Fund III, STAR Market standards, directed institutional demand — will continue to fund domestic chip champions. The question is not whether China will keep financing its chip industry. It is whether the public market can keep pricing those companies as if profitability were optional. The answer will be written in the names that list after Enflame: XPHOR, the silicon-photonics firm accepted for a 2.43 billion yuan STAR offering; InnoGrit, the storage-controller developer greenlit this year; and others in the pipeline.
The central tension is now visible in a single line. Enflame's valuation is 61.8 times sales with no earnings; CXMT's first-day pop was 466% on a company that only recently turned profitable. The market is being asked to fund a national project and price it as a public equity at the same time. Those two jobs do not always pay the same return.
China's chip founders are billionaires because the state built a market that rewards strategy ahead of earnings. The bet is that earnings will follow. Until they do, the wealth is real on paper — and the test is whether paper is enough.
Explore more exclusive insights at nextfin.ai.
