{"id":23298,"date":"2026-08-13T10:19:25","date_gmt":"2026-08-13T10:19:25","guid":{"rendered":"https:\/\/scannn.com\/the-deepseek-thesis-by-irene-zhang\/"},"modified":"2026-08-13T10:19:25","modified_gmt":"2026-08-13T10:19:25","slug":"the-deepseek-thesis-by-irene-zhang","status":"publish","type":"post","link":"https:\/\/scannn.com\/lv\/the-deepseek-thesis-by-irene-zhang\/","title":{"rendered":"The DeepSeek Thesis - by Irene Zhang"},"content":{"rendered":"\n<div dir=\"auto\">\n<p><span>What does Liang Wenfeng want? The CEO of DeepSeek is now <\/span><a href=\"https:\/\/www.bloomberg.com\/news\/articles\/2026-07-14\/deepseek-s-liang-tops-amodei-and-brockman-as-richest-ai-founder\"><span>richer<\/span><\/a><span> than either Dario Amodei or Sam Altman. Backed by his careful watch and the immense wealth of his hedge fund, a little-known group of researchers in Hangzhou has managed to consistently command the world\u2019s attention \u2014 and they keep giving their work away for free. The motivations behind Liang\u2019s theory of action have puzzled China+AI watchers since the beginning.<\/span><\/p>\n<p><span>In late July, leaked minutes from a four-hour meeting between Liang and investors circulated around the internet. Like everyone else, we\u2019ve been pouring over the minutes to understand Liang the CEO, DeepSeek the company, and Liang the man. Liang comes off as a genuinely unusual character among China\u2019s tech elite: a person who sincerely holds a specific and highly personalized worldview, if not an entire ideology. Whereas others in his walk of life are highly attuned to commercial trends and political winds, he positions himself as squarely focused on realizing his vision for technology \u2014 and it\u2019s hard to doubt his conviction when he has invested so much of his literal worth into this quest.<\/span><\/p>\n<p><span>His is a confidently singular thesis: that there is an inevitable causal relationship between automated learning and generalized intelligence, and that the pursuit of such advanced machine intelligence is the only problem worth solving right now. Being Chinese means working within the constraints of current hardware limitations, but DeepSeek frames their own ambitions on a much farther horizon. Whether or not they can, of course, might not be up to them.<\/span><\/p>\n<p><span>At ChinaTalk, we\u2019ve been <\/span><a href=\"https:\/\/www.chinatalk.media\/p\/deepseek-v4\"><span>covering<\/span><\/a><span> the DeepSeek story since <\/span><a href=\"https:\/\/www.chinatalk.media\/p\/deepseek-ceo-interview-with-chinas\"><span>before R1<\/span><\/a><span>. The following are our thoughts and observations after reading the leaked minutes. We get into:<\/span><\/p>\n<ul>\n<li>\n<p><span>How Chinese open models work as a business model;<\/span><\/p>\n<\/li>\n<li>\n<p><span>Why, according to Liang, learning is the path to AGI;<\/span><\/p>\n<\/li>\n<li>\n<p><span>What China\u2019s political economy looks like from the vantage point of Hangzhou;<\/span><\/p>\n<\/li>\n<li>\n<p><span>How DeepSeek is managed differently;<\/span><\/p>\n<\/li>\n<li>\n<p><span>And what the Liang Wenfeng-Demis Hassabis comparison reveals and misses.<\/span><\/p>\n<\/li>\n<\/ul>\n<p><span>If the weights are open, how exactly does DeepSeek make money off of its models? Liang told investors that the lab\u2019s biggest source of revenue comes from business customers. When R1 came out in 2025, there was a wave of Chinese businesses and government entities connecting DeepSeek to their internal systems. While much of that was a fad, some of those corporate customers apparently stuck. Since all their models since R1 have been released openly under the MIT license, DeepSeek does not earn revenue from on-premise deployments by corporate entities, so those actual corporate customers must be paying for API tokens.<\/span><\/p>\n<p><span>The DeepSeek team published an <\/span><a href=\"https:\/\/zhuanlan.zhihu.com\/p\/27181462601\"><span>analysis<\/span><\/a><span> in February 2025 showing that on an average day, R1\u2019s API earned the lab US$562,027, at a 545% profit margin. Fast forward to now, and the hypothetical figure Liang apparently gave in the investor meeting for DeepSeek\u2019s enterprise-end revenue this year is in the hundreds of millions of US dollars. With enough growth in this area, he figured, DeepSeek could even turn a net profit soon, paving the way for a successful IPO. Failing that, and API sales to individual consumers have room for growth as well.<\/span><\/p>\n<p><span>That being said, Liang has little interest in consumers. He confesses that at one point last year, the lab even considered sunsetting its consumer products given how little effort went into maintaining them. But DeepSeek\u2019s chatbot and individual API users were incredibly loyal, apparently, so the team ultimately decided to keep the lights on. (I wonder how many of these are <\/span><a href=\"https:\/\/www.chinatalk.media\/p\/ai-roleplay-a-deep-dive\"><span>roleplayers<\/span><\/a><span>\u2026)<\/span><\/p>\n<div class=\"captioned-image-container\">\n<figure><a target=\"_blank\" href=\"https:\/\/substackcdn.com\/image\/fetch\/$s_!5mzB!,f_auto,q_auto:good,fl_progressive:steep\/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F6499a828-96ec-4b70-9085-bde52056aa92_1179x1161.jpeg\" data-component-name=\"Image2ToDOM\" class=\"image-link image2 is-viewable-img can-restack\"><\/p>\n<div class=\"image2-inset\"><picture><source type=\"image\/webp\" srcset=\"https:\/\/substackcdn.com\/image\/fetch\/$s_!5mzB!,w_424,c_limit,f_webp,q_auto:good,fl_progressive:steep\/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F6499a828-96ec-4b70-9085-bde52056aa92_1179x1161.jpeg 424w, https:\/\/substackcdn.com\/image\/fetch\/$s_!5mzB!,w_848,c_limit,f_webp,q_auto:good,fl_progressive:steep\/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F6499a828-96ec-4b70-9085-bde52056aa92_1179x1161.jpeg 848w, https:\/\/substackcdn.com\/image\/fetch\/$s_!5mzB!,w_1272,c_limit,f_webp,q_auto:good,fl_progressive:steep\/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F6499a828-96ec-4b70-9085-bde52056aa92_1179x1161.jpeg 1272w, https:\/\/substackcdn.com\/image\/fetch\/$s_!5mzB!,w_1456,c_limit,f_webp,q_auto:good,fl_progressive:steep\/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F6499a828-96ec-4b70-9085-bde52056aa92_1179x1161.jpeg 1456w\" sizes=\"100vw\"\/><\/picture><\/div>\n<p><\/a><figcaption class=\"image-caption\">One of the many memes that circulated around the Chinese internet after Liang\u2019s comments leaked; this one was posted to Xiaohongshu by user @lunerpine. The caption reads, \u201cDon\u2019t know what the point of users is\u2026 I\u2019ll just feed them for now.\u201d<\/figcaption><\/figure>\n<\/div>\n<p><span>Minor problems like keeping users around do not concern Liang. As long as the upper limits of machine intelligence are still out of view, there is more (technological, societal, and literal business) value in pursuing that than there is in building business models based on what is available today. More capable models easily pull the rug out from under competitors, especially when the latter have fallen into path dependencies based on older technology. Once truly generalized and convincingly superior intelligence arrives, creating to-C and to-B products will be trivial. Before then, the vast majority of DeepSeek\u2019s resources will go into probing the upper limits of intelligence \u2014 and nothing else.<\/span><\/p>\n<p><span>Liang describes this work as the \u201cmain quest\u201d of AGI, which, to him, covers research in the realm of general-purpose intelligence, agents, chain-of-thought, etc. He is careful not to commit to specific subfields or theories, but does explicitly exclude some trajectories. World models, for example, are to him \u201cirrelevant\u201d for the pursuit of higher levels of intelligence, even though he thinks embodied AI is largely inevitable. He argues that once AI research is based on \u201cself-iteration,\u201d these systems will then solve for embodied intelligence in the physical world. Before then, however, he believes the most important problem right now is learning.<\/span><\/p>\n<p><span>AI training today relies heavily on high-quality, labelled data being fed into models \u2014 not autonomous and truly continuous learning. Liang cautions investors not to think of \u201clearning\u201d as a sub-technology the way they do with \u201cagents,\u201d but instead to think of it as a problem to be solved. (Indeed, one takeaway from reading Liang in his own words is that despite being a finance veteran, he seems to see the world as an unfolding kaleidoscope of puzzles the way an academic researcher does.) The path to AGI, in his eyes, runs through mechanisms that allow models to keep acquiring knowledge. He is quite honest that he and his team don\u2019t know what that path looks like yet, but he\u2019s asking investors to bet on their clarity of thought.<\/span><\/p>\n<p><span>Throughout the conversation, Liang seems to implicitly assume that future AI systems \u2014 and eventually AGI \u2014 will be open. While America will maintain its capability advantage for now, China, in his eyes, will play the role of token factory at global scale, pushing the price of intelligence down as it did for countless other industries during its manufacturing boom.<\/span><\/p>\n<p><span>Liang appears to have somewhat settled into the national-champion role. Phrases like \u201chistoric mission\u201d slip through when he discusses how the hardware chokehold China faces will be eroded and finally dismantled, as if the forces of history make that inevitable. Liang expects Nvidia\u2019s CUDA moat to erode, and \u2014 with limited specifics \u2014 expressed cautious optimism about training on Huawei chips. In fact, he presents DeepSeek\u2019s relationship with Huawei as cordial and regards working with domestic GPUs as an inevitability. Huawei apparently allocated 16,000 Ascend 950 GPUs to DeepSeek, which is fewer than what it sold to \u201cbigger internet companies\u201d (probably ByteDance, Alibaba, and Tencent). Revealingly, Liang remarked that these competitors need Huawei chips more than DeepSeek does because DeepSeek can \u201cacquire some noncompliant chips,\u201d and that DeepSeek\u2019s main rationale for spending on Huawei is to support a domestic hardware ecosystem.<\/span><\/p>\n<p><span>Liang does not want to be an out-and-proud part of China, Inc. He knows, however, that Chinese domestic integration is inevitable, and sees value in aligning with Beijing on tech sovereignty. The trouble is convincing investors that they should really feel so optimistic about model training on domestic GPUs \u2014 it\u2019s the topic that collected the most number of questions during the Q&amp;A.<\/span><\/p>\n<p><span>Liang clearly takes great pride in the company he\u2019s built. To investors, he pitched a story of a team that commits to its work not only because of the substance or the mission, but also because of a culture that\u2019s rare in China\u2019s tech industry. DeepSeek researchers, apparently, rarely work overtime. Management strives to ensure that \u201cmandatory\u201d tasks \u2014 those based on central goals decided collectively \u2014 do not take up more than half of employees\u2019 time at work. The other half is entirely self-directed, and employees are supported in pursuing whatever research projects interest them as long as there is compute available.<\/span><\/p>\n<p><span>Of course, Liang is not doing this entirely out of the goodness of his heart (though he does seem to have a refreshingly normal view of human performance, having once <\/span><a href=\"https:\/\/www.economist.com\/business\/2026\/07\/19\/chinas-mysterious-new-billionaires-are-conquering-the-world\"><span>noted<\/span><\/a><span> that he believes a person can only concentrate for six to eight hours a day). In the meeting, he argued that a relaxed environment is a necessity for research. Good research ideas come from exploring idle curiosity, and there would be no room for such curiosity if researchers were constantly preoccupied with pressure. To maintain such an environment, Liang focuses on \u201csaying no\u201d to projects that do not relate to their main mission; he told investors that this is why they intentionally did not make improvements to many existing products.<\/span><\/p>\n<p><span>Will a new theory of technology and productivity finally defeat the dreaded 996? Rather than jumping onto every vertical and rushing to be first, as is the norm across much of China\u2019s internet industry, Liang is betting that focus and discipline will win the day in the AI chapter. Unlikely in Silicon Valley, where the AI gold rush is unleashing <\/span><a href=\"https:\/\/www.youtube.com\/watch?v=VgkXOE-V-fc&amp;vl=en-US\"><span>extremely workaholic<\/span><\/a><span> founders and pushing even longer hours onto employees, Liang doesn\u2019t sound like he is racing against time. (He\u2019s certainly not ending up in the permanent underclass personally.) In his eyes, it seems, the horizons for machine intelligence are still so wide that there is much more work to be done before anyone should panic. It\u2019s in the interest of the follower, after all, to hope that the race remains far from over. But maybe Liang is also less worried about the consequences of such a race: throughout the four-hour conversation, the dangers of potential AGI never come up. The pursuit of it seems inherently worthy to Liang.<\/span><\/p>\n<p><span>Liang is inspired by the history of Bell Labs but does not see DeepSeek as an exact replica. When investors asked about the tension between DeepSeek\u2019s research idealism and the urgency of constructing a sustainable business, Liang argued that missions do not cost good companies revenue. Rather, worthy missions make for even stronger paths to commercialization. In other words, not only is research possible inside for-profit companies, but sometimes better outcomes happen when there is capital involved.<\/span><\/p>\n<p><span>There\u2019s another AI CEO who arrived at a similar conclusion \u2014  Demis Hassabis. Sebastian Mallaby\u2019s biography of Hassabis, <\/span><em><a href=\"https:\/\/a.co\/d\/00DILtJ7\"><span>The Infinity Machine<\/span><\/a><\/em><span>, documents how DeepMind became a Google subsidiary. Former Google CEO Larry Page presented Hassabis, who resented business development work, two future scenarios: either spend the prime of his career gathering Google-level resources for the real work DeepMind hopes to accomplish, or do that work right now on Google\u2019s back. To Hassabis, the choice was obvious.<\/span><\/p>\n<p><span>But Demis\u2019s bet did not turn out the way he wanted. Google DeepMind became much more entwined in Google\u2019s corporate labyrinth than he expected. He complained of becoming just another employee from the perspective of Mountain View, as he was no longer the sole deciding vote steering what DeepMind produced. Unlike Liang, who regards harnessing commercial incentives for \u201cpure\u201d AGI development as entirely possible, Mallaby describes Hassabis as disappointed in what transpired, even though he remains convinced of the necessity of keeping powerful technology in the hands of those who understand it most.<\/span><\/p>\n<p><span>On August 5th, all this tension ripped wide open. Hassabis has resigned from Google and will pivot to working on AI-assisted drug discovery at <\/span><span>Isomorphic Labs, as well as researching \u201csocietal impacts of AGI\u201d per a <\/span><a href=\"https:\/\/www.reuters.com\/business\/google-shakes-up-ai-leadership-deepmind-chief-shifts-role-2026-08-05\/\">Google spokesperson<\/a><span>. The siren call of the mission pulled him away from a disheartening situation \u2014 but also away from his life\u2019s work at DeepMind.<\/span><\/p>\n<p><span>Hassabis should certainly not be jealous of Liang\u2019s personal predicament. While Hassabis was merely constrained by the complex politics of a world-famous company, Liang\u2019s firm now symbolizes the US-China AI race. To shield his company from the eye of the storm, Liang takes pains to shape the domestic narrative around DeepSeek. There is to be no heroism or glorification of specific researchers; he repeatedly emphasized that DeepSeek is made up of \u201cregular people\u201d whose secret sauce is teamwork. He mentions at one point that the government \u201cwon\u2019t give [them] a cent\u201d should the company fail. This shouldn\u2019t be taken literally given the asymmetric relationship between the Communist Party\u2019s coercive potential and a small (if well-funded) lab; it was probably to assure his audience of investors that sound financial management won\u2019t take a backseat as DeepSeek pours more of itself into AGI.<\/span><\/p>\n<p><span>After leaks of this very meeting spiraled across the internet, Liang was reportedly so <\/span><a href=\"https:\/\/seekingalpha.com\/news\/4618437-deepseek-is-said-to-pause-new-funding-round-as-ipo-plans-advance\"><span>furious<\/span><\/a><span> that he paused a new funding round and pushed back IPO plans once more. More than anything, he demands complete trust that does not dilute his exact vision \u2014 and whoever leaked this clearly does not fully trust the plan. The question is: who, among the audience that day, had reasons to want to poke holes in Liang\u2019s vision?<\/span><\/p>\n<\/div>\n<p><a href=\"https:\/\/www.chinatalk.media\/p\/the-deepseek-thesis?utm_source=tldrnewsletter\">Source link <\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>What does Liang Wenfeng want? The CEO of DeepSeek is now richer than either Dario Amodei or Sam Altman. Backed by his careful watch and the immense wealth of his hedge fund, a little-known group of researchers in Hangzhou has managed to consistently command the world\u2019s attention \u2014 and they keep giving their work away [&hellip;]<\/p>\n","protected":false},"author":16,"featured_media":23299,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[143],"tags":[],"class_list":["post-23298","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-ai"],"_links":{"self":[{"href":"https:\/\/scannn.com\/lv\/wp-json\/wp\/v2\/posts\/23298","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/scannn.com\/lv\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/scannn.com\/lv\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/scannn.com\/lv\/wp-json\/wp\/v2\/users\/16"}],"replies":[{"embeddable":true,"href":"https:\/\/scannn.com\/lv\/wp-json\/wp\/v2\/comments?post=23298"}],"version-history":[{"count":0,"href":"https:\/\/scannn.com\/lv\/wp-json\/wp\/v2\/posts\/23298\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/scannn.com\/lv\/wp-json\/wp\/v2\/media\/23299"}],"wp:attachment":[{"href":"https:\/\/scannn.com\/lv\/wp-json\/wp\/v2\/media?parent=23298"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/scannn.com\/lv\/wp-json\/wp\/v2\/categories?post=23298"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/scannn.com\/lv\/wp-json\/wp\/v2\/tags?post=23298"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}