![DeepSeek: The 7 Billion Dollar Bet](https://cdn.slatesource.com/e/7/5/e750355d-4dfd-4aec-b53e-d23d8bac7e11.webp) # DeepSeek: The 7 Billion Dollar Bet - [Made in Slatesource](https://slatesource.com/@steph/deepseek-the-7-billion-dollar-bet) - By [Steph](https://slatesource.com/@steph) - Created on Jul 23, 2026 > Money has never been the problem for us. Bans on shipments of advanced chips are the problem. ## DeepSeekDeepSeek: The LabLab That ChangedChanged EverythingEverything In JanuaryJanuary 20252025, a smallsmall ChineseChinese AI lablab releasedreleased an open-sourceopen-source reasoningreasoning modelmodel that erasederased $593593 billionbillion from Nvidia'sNvidia's marketmarket capcap in a singlesingle dayday. The modelmodel matchedmatched OpenAI'sOpenAI's bestbest for a reportedreported trainingtraining costcost of $5.576576 millionmillion. MarcMarc AndreessenAndreessen calledcalled it "AI'sAI's SputnikSputnik momentmoment." For 18 monthsmonths afterafter that shockshock, DeepSeekDeepSeek refusedrefused everyevery investorinvestor, everyevery chequecheque, everyevery overtureoverture from China'sChina's biggestbiggest techtech firmsfirms. Then, in JuneJune 20262026, it openedopened the doordoor and raisedraised $7.4 billionbillion in oneone roundround at a $52 billionbillion valuationvaluation. The AI racerace just tiltedtilted againagain. First external funding round $7.4 billion (50 billion yuan), June 2026 Post-money valuation approximately $52 billion Second round in talks targeting $71 to $74 billion pre-money Employees approximately 300, plans to double to 400 Founded July 2023, Hangzhou, China IPO target Shanghai Star Market, possible debut Q2 2027 ## The RoundRound: Who WroteWrote the ChequesCheques The investorinvestor listlist readsreads likelike a who'swho's who of ChineseChinese technologytechnology and industryindustry. TencentTencent ledled with 10 billionbillion yuanyuan. CATLCATL, the world'sworld's largestlargest EV batterybattery makermaker, contributedcontributed 5 billionbillion yuanyuan, a betbet that AI and energyenergy storagestorage are convergingconverging. JD.comcom, NetEaseNetEase, and IDGIDG CapitalCapital eacheach putput in 3 billionbillion yuanyuan. China'sChina's nationalnational AI industryindustry fundfund investedinvested 1 billionbillion yuanyuan directlydirectly into DeepSeek'sDeepSeek's corecore entityentity and, unusuallyunusually, retainedretained votingvoting rightsrights. EveryEvery otherother externalexternal investorinvestor acceptedaccepted a lock-uplock-up of up to fivefive yearsyears with no votesvotes. LiangLiang WenfengWenfeng himselfhimself contributedcontributed 20 billionbillion yuanyuan of his ownown moneymoney, nearlynearly halfhalf the roundround, cementingcementing his 78 percentpercent equityequity stakestake. # DeepSeek trained a world-class AI model for $5.6 million while US rivals spend \\ billions. Now it has $7.4 billion more. Does algorithmic efficiency plus capital make it unstoppable, or does the West's chip advantage still matter? No comments yet. Be the first. Log in to comment R1 training cost (reported) $5.576 million vs billions for US rivals AIME 2024 math benchmark 79.8 percent (vs OpenAI o1: 79.2 percent) MATH-500 score 97.3 percent (vs o1: 96.4 percent) API input price $0.28 per million tokens (vs GPT-4o: $2.50, roughly 9x cheaper) China market share (early 2026) 89 percent of AI API and app usage Nvidia market cap erased on R1 launch day $593 billion (January 27, 2025) The Architecture Edge DeepSeek-R1 uses a Mixture-of-Experts design: 671 billion total parameters but only 37 billion activate per inference pass. That keeps compute proportional to a much smaller model. The technique is not new, but DeepSeek applied it at a scale and cost efficiency that rivals had not. DeepSeek-V4, released April 2026, added a native 1 million-token context window and training on 32 trillion tokens. The Control Structure External investors face a lock-up of up to five years and hold zero voting rights. The sole exception is China's national AI fund, which retained a vote. Liang Wenfeng holds 78 percent equity through a special structure. One analyst noted: "When decision-making power is this concentrated, the cost of correcting a strategic obsession is extremely high." Liang told investors: open-source research comes before short-term commercialisation. Always. ## What the MoneyMoney Is For The immediateimmediate prioritiespriorities are threefoldthreefold. FirstFirst, computingcomputing capacitycapacity: DeepSeek'sDeepSeek's internalinternal chipchip stockpilestockpile remainsremains a genuinegenuine bottleneckbottleneck eveneven afterafter recentrecent GPUGPU buysbuys, and LiangLiang has toldtold the ChineseChinese premierpremier as muchmuch. SecondSecond, the teamteam: headcountheadcount will roughlyroughly doubledouble from 300300 to 400400, a tinytiny numbernumber by US frontier-labfrontier-lab standardsstandards. ThirdThird, a longer-termlonger-term betbet on buildingbuilding its ownown AI chipchip, reducingreducing dependencedependence on NvidiaNvidia hardwarehardware subjectsubject to US exportexport controlscontrols. The IPOIPO on Shanghai'sShanghai's StarStar MarketMarket, expectedexpected no soonersooner than Q2 20272027, wouldwould givegive DeepSeekDeepSeek accessaccess to publicpublic capitalcapital marketsmarkets and a currencycurrency for futurefuture dealsdeals. The Export Control Shadow US export restrictions on advanced chips have directly shaped DeepSeek's story at every turn. R1 was trained on Nvidia H800 chips, which were not restricted until October 2023. The House Select Committee on the CCP has accused DeepSeek of stockpiling chips before controls took effect and recommends expanding restrictions further. A recurring irony in analyst reports: the pressure to innovate around chip scarcity is precisely what drove the algorithmic breakthroughs that [DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning (original paper)](https://arxiv.org/abs/2501.12948?utm_source=slatesource)