DeepSeek Accelerates STAR Market IPO: The $50B Valuation, CITIC Sponsorship, and What It Means for Chinese AI Capital
On September 9, 2026, Reuters reported that DeepSeek has selected CITIC Securities as its sponsor for a Shanghai STAR Market listing, with plans to submit IPO application materials before the end of 2026 and list on the board in 2027. If completed, it would be the first independent general-purpose large-model company to land on the A-share market — and the most consequential capital event in Chinese AI since the industry’s emergence.
From Self-Funded Rebel to $50B+ Valuation: The Five-Month Capital Blitz
The speed of DeepSeek’s capital transformation is unprecedented. Consider the timeline:
| Date | Event |
|---|---|
| April 2026 | DeepSeek launches its first external funding round |
| June 2026 | Closes the round at RMB 51B (~$7.2B), post-money valuation ~$50B |
| July 2026 | Launches second round; target valuation climbs to RMB 500B (~$71B) |
| August 2026 | Second round oversubscribed; IPO preparation accelerates |
| September 2026 | CITIC Securities confirmed as sponsor |
Just five months ago, DeepSeek was the self-funded rebel of Chinese AI — Liang Wenfeng’s High-Flyer Quant hedge fund had bankrolled the lab since its founding, and DeepSeek marketed itself as the model company that did not need venture capital. Today, it is on track to deliver the largest AI IPO in Chinese history.
The Financing Breakdown: Who Is Writing the Checks
The June 2026 round already sets records:
- Liang Wenfeng (personal): RMB 20B — largest single investor
- Tencent: RMB 10B
- CATL (Contemporary Amperex Technology): RMB 5B
- NetEase, JD.com, Monolith Capital, IDG Capital: RMB 3B each
- National AI Industry Investment Fund: ~RMB 1B strategic investment
The second round brings in new institutional and industrial capital: SMIC Private Equity, Boyu Capital, CPE Yuanfeng, and state-backed Hefei investment platforms. Tencent, CATL, and IDG are all doubling down.
The combined two-round total could exceed RMB 100B (~$14B), making it the largest AI fundraising event outside of the United States.
Why the STAR Market, and Why Now
Three forces converged to push DeepSeek toward an accelerated IPO timeline:
1. Compute Capital Outgrew the Hedge Fund Balance Sheet
Liang Wenfeng’s High-Flyer Quant, while successful, could no longer absorb the compute budget required to stay at the frontier. DeepSeek’s 2025 infrastructure spend was approximately RMB 1.2B. In the first seven months of 2026 alone, it rose to ~RMB 11B — nearly a 10× increase within nine months. The Ulanqab gigawatt-scale compute center being built in Inner Mongolia will require continued capital infusions that no single investor, even one as wealthy as Liang, can cover.
2. The STAR Market Policy Window Opened in June 2026
On June 17, 2026, the Shanghai Stock Exchange relaxed its Fifth Listing Standard, explicitly creating a pathway for unprofitable hard-tech companies — including AI model firms — to list on the STAR Market. The policy change removed a key structural barrier. DeepSeek is the first major AI lab to step through that door.
3. Talent Retention Demanded Liquidity
Multiple core DeepSeek researchers left earlier in 2026 for Tencent, Xiaomi, and ByteDance. The departures were tied to illiquid stock options — engineers cannot wait 3–4 years for a liquidity event. An IPO creates the exit path that keeps top talent in-house.
Liang Wenfeng’s Control Structure: Tight as a Founder Gets
Remarkably, Liang Wenfeng is using the capital infusion to strengthen, not dilute, his control:
- Before the first round, Liang increased his direct shareholding to 34% via industrial and commercial capital increases
- Combined with indirect holdings, his total control exceeds 84%
- Voting rights approach 100% — almost all external investors hold shares through limited partnership platforms Liang manages
- External shareholders lack voting rights and face a five-year lock-up (reduced to three for new round investors)
This structure isolates DeepSeek from external capital interference — a deliberate choice for a founder who has seen what happens when Western AI labs capitulate to investor pressure.
The Revenue Picture: Growing Fast, Still Losing Money
For a company approaching a $50B+ valuation, DeepSeek’s financials are young:
- 2026 H1–H7 revenue: RMB 475M (~$66M) — already 10× 2025’s full-year revenue
- Net loss (Jan–July 2026): RMB 715M (~$100M)
- Gross margin: ~44.6% — strong for a model company, driven primarily by API business
The revenue multiple is extraordinary. At $71B pre-money valuation on roughly $500M annualized revenue, DeepSeek trades at ~148× P/S. By comparison:
| Company | P/S Ratio |
|---|---|
| DeepSeek (pre-round) | ~148× |
| OpenAI | ~65× |
| Anthropic | ~21× |
| Zhipu AI (Hong Kong-listed) | ~957× |
| Kweichow Moutai | ~10× |
This is not a valuation justified by financial metrics. It is a valuation justified by scarcity — one of a small handful of independent AI labs producing frontier open-weight reasoning models, operating in a jurisdictional bubble where U.S. competitors cannot access its market.
What This Means for DeepThink
The DeepThink reasoning engine that powers DeepSeek’s V4.1 Flash and V4 Pro models is the core intellectual asset behind this IPO. For users and operators of the DeepThink ecosystem, the IPO affects three things directly:
- Compute capacity increases. The IPO proceeds fund the Ulanqab center and domestic inference chip development, meaning more GPU hours for production DeepThink workloads.
- API pricing stabilizes. A public company needs revenue predictability. The volatility of 2026’s pricing war may settle into a more predictable tiering model, benefiting enterprise users.
- Regulatory alignment. The STAR Market listing puts DeepThink — and Chinese AI generally — on a regulatory footing that parallels, rather than conflicts with, global AI governance frameworks.
The Risk: A $2.5T Cap, But What’s the Multiple?
Analysts project a post-listing market cap between RMB 1.5T and RMB 2.5T ($210B–$350B). That would make DeepSeek one of the top 15 public companies in China by market cap, and the largest AI company on A-shares.
But multiple questions remain unresolved:
- Will the STAR Market accept a company that is still net-loss-making?
- Can DeepSeek’s revenue growth justify the valuation, or will it become the AI equivalent of a high-multiple growth stock that disappoints?
- What happens to the open-weight strategy once DeepSeek is accountable to public investors who want monetization, not community goodwill?
Looking Ahead
The DeepSeek IPO is more than a liquidity event. It is the moment when Chinese frontier AI formally enters the global capital stage. When the company lists in 2027, it will set a public multiple against which every private AI lab valuation — in both China and the West — will be compared.
For the DeepThink reasoning community, this means more compute, more stable APIs, and a clearer sense of which models are here to stay. The V4.1 Flash pricing is not just competitive against V4 Pro. It is a preview of what a well-capitalized, publicly accountable DeepThink ecosystem can deliver.
The IPO file itself will tell a story that matters far beyond A-share investors. When CITIC Securities submits it, the DeepThink reasoning engine will be valued not just on benchmarks, but on its ability to move markets — and that is a watershed moment for the entire industry.