DeepSeek funding timeline: from "no fundraising" to a $70B valuation in four months
Every second-round figure below — amount, valuation, timeline — comes from anonymous dealmakers cited by Chinese financial media, not from an official DeepSeek statement. Terms could still shift before signing.
April 2026 — ownership reset: A filing shows registered capital rose and founder Liang Wenfeng personally subscribed, lifting his direct stake from 1% to 34%. Combined with an entity he controls, total control reached roughly 84.29%. The same month DeepSeek opened its first external round and previewed V4.
June 2026 — Round 1 closed: Roughly 50 billion yuan (~$7.4 billion) at a post-money valuation above 350 billion yuan ($52–59 billion across sources) — the largest first-round raise in Chinese AI history. Liang personally put in 20 billion yuan; Tencent 10 billion, CATL 5 billion, plus JD.com, NetEase, IDG, and China's National AI Industry Investment Fund.
July 14–17 — IPO prep + Round 2 talks: Outlets reported STAR Market (Shanghai) IPO prep and a second round at roughly $71 billion pre-money (~480 billion yuan), about 37% above Round 1 post-money. ARR of $400–500 million (mostly API tokens) surfaced for the first time.
July 25–26 — abrupt pause: Bloomberg and others said Liang was unhappy that closed-door investor remarks had circulated online; some standby investors were told to hold off on signing.
August 4–5 — restart: Caijing-cited dealmakers said talks resumed at ~50 billion yuan and ~500 billion yuan pre-money (~$70 billion), with signing expected in late August. Both sides want the round kept low-profile.
Note: From an ~$10B narrative around the April open, to ~$52B after June close, to ~$70B pre-money now, reported valuations roughly 7x in under four months — all still media-sourced.
DeepSeek Round 1 vs Round 2: the numbers at a glance
Two rounds, same check size, stepped-up valuation. If Round 2 closes, combined proceeds exceed 100 billion yuan (~$14B) in under five months.
| Round 1 (closed) | Round 2 (in talks) | |
|---|---|---|
| Talks opened | April 2026 | Restarted mid-July, paused, restarted Aug 4–5 |
| Expected/actual close | June 2026 | Late August 2026 (planned) |
| Amount | ~50B yuan (~$7.4B) | Target ~50B yuan (~$7B) |
| Valuation basis | Post-money >350B yuan | Pre-money ~500B yuan (~$70B) |
| Valuation increase | — | ~+43% vs Round 1 |
| Key backers | National AI Fund, Tencent (10B yuan), CATL (5B yuan), JD.com, NetEase, IDG, Loyal Valley, Shixiang | Round-1 runners-up + some existing backers adding |
| Combined if Round 2 closes | — | Over 100B yuan (~$14B) in under 5 months |
| Metric | Value | Note |
|---|---|---|
| ARR | ~$400–500 million | Mostly API tokens; media-sourced, not official |
| Gross margin | Reportedly >50% | Unaudited |
| Implied P/S | ~140–150x | Vs OpenAI ~65x and Anthropic ~21x (dealmaker estimates) |
| MAU | 100M+ (external reports) | Methodology undisclosed |
"Pricing a foundation-model company is fundamentally an options bet, not a cash-flow valuation." — a dealmaker quoted in Chinese coverage of the 140–150x multiple.
Inside the deal: compute bills, missing votes, and a 148x price-to-sales ratio
Three threads explain the math better than the headline number alone.
The real bill is compute. After Round 1, DeepSeek said it would double headcount in data-center and AI-agent teams; Reuters reported hiring for in-house AI inference chips. Analysts estimate ~7 of every 10 billion yuan raised goes into chips, data centers, bandwidth, and liquid cooling. Cadence is a race against compute buildout. See also our DeepSeek custom-chip brief.
Most investors don't get a vote. Round 1 capital largely entered via an LP controlled by Liang — no voting rights, five-year lock-up. The exception: China's National AI Industry Investment Fund invested directly with votes and no lock-up. Liang's ~84% control holds; Forbes and others have flagged governance and state-alignment questions.
148x P/S is an options bet — or a red flag. At ~$70B pre-money against $400–500M ARR, implied P/S sits ~140–150x versus OpenAI ~65x and Anthropic ~21x. Investors are pricing infrastructure-level upside in China's compute stack and enterprise agents, not today's cash flows.
Caveat: Deal size, valuation, and cap-table details come from anonymous-sourced reporting (Caijing, Reuters, Bloomberg, Forbes, among others). DeepSeek has not officially confirmed Round 2 terms.
How DeepSeek stacks up against Moonshot, Zhipu, MiniMax — and a six-step diligence runbook
Still-private leaders DeepSeek and Moonshot both carry ~140–150x P/S multiples, above what listed Zhipu and MiniMax trade at in the secondary market.
| Company | Listing | Latest valuation / market cap | Reported ARR | Recent funding pace |
|---|---|---|---|---|
| DeepSeek | Private; STAR IPO prep | ~500B yuan pre-money (~$70B, in talks) | ~$400–500M | 2 rounds in 4 months; targeting >$14B combined |
| Moonshot AI (Kimi) | Private | ~$20B (May 2026); later talks cited ~$30B | ~$200M | 4 rounds in 6 months; ~$3.9B total |
| Zhipu AI (Z.ai) | Listed (HK) | ~350B yuan market cap (May 2026) | Undisclosed | ~8.3B yuan pre-IPO |
| MiniMax | Listed (HK) | ~210B yuan market cap (May 2026) | Undisclosed | ~11B yuan pre-IPO |
Six-step diligence runbook:
Label closed vs in talks. Treat the $70B pre-money and $7B target as negotiation chatter until a formal announcement. Stamp internal memos "media-sourced / unconfirmed."
Back into the multiple. Divide ~$70B by $400–500M ARR yourself; compare to OpenAI ~65x and Anthropic ~21x so you know whether you are underwriting cash flows or options.
Map voting rights. Round 1 outsiders mostly sat in a founder-controlled LP with no votes and a five-year lock-up; only the National AI Fund got direct votes. Capital ≠ control.
Translate raise size into compute spend. The ~70% of each 10B yuan into chips/DCs/bandwidth frame is a useful stress test for whether Round 2 still funds infrastructure race conditions.
Track the STAR Market fifth standard. On June 17, 2026, Shanghai expanded the unprofitable-AI listing path. DeepSeek reportedly aims to file by end-2026 for a 2027 listing — put IPO and fundraising on one Gantt chart.
Keep an engineering dual track. Funding headlines do not replace a stable Apple Silicon host for agents, CI, and local inference. Laptops that sleep and Linux VPS boxes without Xcode/Metal erase cheap-API gains. Baseline on MESHLAUNCH cloud Mac rental; for sector context see our AI funding supercycle brief.
Controversy, STAR Market rule changes, and why the global AI funding race matters
Leaked closed-door transcript stalled the deal. The July pause reportedly followed Liang's frustration that first-round investor remarks spread online — a reminder that scale makes a low profile harder.
Voting-rights structure draws scrutiny. Most external investors have no vote and a five-year lock-up; only the state-backed National AI Fund gets direct votes with no lock-up. Governance and state-influence questions remain unresolved.
Valuation-to-revenue gap is live. A 140–150x P/S is extreme even versus top SaaS at 30–50x. Whether it holds depends on converting technical lead into scaled enterprise revenue after a STAR listing — still untested.
Why it matters: STAR Market's fifth listing standard now covers AI companies that need not be profitable or heavily revenue-bearing. DeepSeek abandoned a five-year "no fundraising, no IPO, no commercialization" stance funded by High-Flyer. Globally, OpenAI was reportedly valued at $300B in 2025 and Anthropic reportedly surpassed it by June 2026 — steep premiums are not China-only. Compute self-reliance (in-house inference chips, owned DCs) explains why modest ARR still forces fast raises under export-control pressure.
Sources: Caijing (via Sina Finance / Wall Street CN), The Standard HK, Forbes, SCMP, Caixin, Reuters, Bloomberg, CIW, 36Kr, TMTPost. Most figures are anonymous-sourced; verify before publishing.
Funding headlines do not replace a production host. Always-on agents, parallel sub-agents, Xcode CI, and Metal inference still need stable Apple Silicon — sleeping laptops and Metal-less VPS boxes burn the savings from cheap model APIs. For 24/7 exclusive Apple Silicon with day/week/month flexibility, MESHLAUNCH Mac Mini cloud rental is usually the stronger production baseline. See the pricing page and help center.
Not yet. As of this writing, the round is still in negotiation, targeting a close by late August 2026. The final amount and terms could differ from what's currently being reported.
The company is funding a rapid buildout of data centers, in-house AI chips, and headcount across its agent and infrastructure teams — capital expenditure that's outpacing what its first raise covered, according to multiple reports.
No. It comes from dealmakers cited anonymously by Chinese financial media (primarily Caijing), not from an official DeepSeek statement, and it could change before any agreement is signed.
Not necessarily — and that's part of the controversy. In the first round, most outside investors received no voting rights and a five-year lock-up, while only China's National AI Industry Investment Fund got direct voting rights.
DeepSeek is reportedly preparing to file for a STAR Market listing in Shanghai by end-2026, targeting a 2027 debut. This private round is institutional-only; international retail access would likely be indirect after listing. For engineering compute today, see the pricing page and help center.