Behind DeepSeek's Hundred-Billion Financing: Zero AI Returns and Bubble Risk

**Key Points**

**1. The 100 billion yuan financing is buying a position in the AI foundation layer**

DeepSeek's raise was oversubscribed to nearly 100 billion yuan; Tencent wants the traffic gateway, CATL wants the power; the plan is to use at least 160,000 Ascend chips for domestic-compute pre-training.

**2. AI has not yet translated into output for non-tech companies**

Apollo says revenue per employee for the S&P 493 has been flat at 450–550, with no rise in margins or productivity; AI spending is highly concentrated, with the top 10% of companies accounting for 99.5% of model-service and inference spending, partly self-circulating VC money.

**3. Enterprise AI projects have a high failure rate, and giants and startups alike are clearing the field**

About 42% of projects are abandoned by their parent companies; OpenAI has cut Sora and Atlas, at least 25 domestic AI applications shut down in 2025, and 392 AI tools worldwide have ceased service.

**4. Autonomous driving has already proven that technical success does not equal commercial success, and embodied intelligence may repeat the bubble**

More than 40 autonomous-driving companies exited between 2019 and 2026, mostly due to financing breakdowns and commercialization difficulties; DeepSeek took a stake in Unitree, and Liang Wenfeng is also set to lead a round in Winverse AI—robots burn more money and have no sustainable model. 2026-10-26 11:46 A 100 billion all-in, yet AI returns are zero—Shanzi Quick Read

This article is from the WeChat public account: Shanzi, author: Rayking629

DeepSeek has raised another 100 billion yuan. This round, which set a domestic record for a single AI financing, was led by Tencent and CATL—one representing the gateway to China's mobile internet, the other representing the battery throne of new-energy vehicles. As soon as the news broke, voices saying "industrial capital is all in on AI" flooded the screen.

But on almost the same day, Apollo chief economist Torsten Slok offered another set of numbers: setting aside the tech giants, AI's financial return for American companies is zero. Revenue per employee, profit margins, total factor productivity—all flat across the 493 non-tech companies in the S&P. The money has not produced results.

On one side is a hundred-billion-level arms race in computing power; on the other is a verdict that "AI's economic return is zero." Only when these two pictures hold at the same time do they constitute the most contradictory moment of capital and technology in 2026.

A hundred billion: whose money is it

In this round of financing, DeepSeek went from a fundraising target of 50 billion yuan all the way to nearly 100 billion yuan in oversubscription, with a pre-money valuation of about 500 billion yuan, absorbing more than 150 billion yuan in capital across two rounds within half a year.

Tencent has made heavy bets in two consecutive rounds (10 billion yuan in Series A, and more in this round), aiming for priority participation in the ecosystem gateway—when WeChat, gaming, and cloud businesses need to connect to model capabilities, it wants to be sitting in the front row. CATL raised its contribution to about 10 billion yuan, but it wants a different account: data-center electricity costs, energy-storage peak-valley arbitrage, direct green-power supply—putting "electricity," the largest cost item in computing power, into its own business map.

One manages traffic, the other manages power. What they are betting on may not be the large model itself, but a position in the "computing foundation"—SMIC Capital (backed by SMIC), Shixi Capital (backed by GigaDevice), and Geely have also squeezed in; demand for computing power from chip manufacturing and the auto industry is forming a positive cycle.

Where the money flows writes the ambition on its face: DeepSeek plans to build a data center in Ulanqab, Inner Mongolia, deploying at least 160,000 Huawei Ascend chips, using domestic computing power for frontier-model pre-training for the first time—Liang Wenfeng himself calls it "the biggest gamble."

The line "what we have never lacked is money" no longer applies. Back then he said the company's problem was the embargo on high-end chips; now the financing logic has become—since the most advanced chips cannot be bought, use massive capital to first smash out a domestic computing ecosystem. Money has become a weapon against the ban.

The economy has not caught this money

Revenue per employee for the S&P 493 (companies excluding the Mag 7) has long been flat in the 450–550 range, while the Mag 7 has climbed toward 800; margins are flat for the former and rising for the latter; productivity growth—whether TFP or output per hour—shows no sign of being lifted by AI.

More striking is the distribution of spending. In model services and inference, the top 10% of enterprises account for 99.5% of spending; among OpenAI and Anthropic customers, the top 10% account for 95.6%; among companies in the top 1% of AI spending, 80.4% are VC-backed, while among median-spending customers, VC-backed firms account for only 13.1%.

A considerable portion of AI consumption is VC money circulating on its own—like the subsidy-burning shared bikes of the past: lively in the app, but no money on the books.

Torsten Slok summed it up in one sentence: mainstream enterprises are beginning to redefine token spending from a "source of growth" to a "cost center that erodes profit." They do not need the latest models, nor are they willing to pay for them.

Foreign media have erected a monument to this frenzy, called the AI Graveyard. According to S&P Global Market Intelligence, about 42% of enterprise AI projects are ultimately abandoned by their parent companies, for reasons no more than insufficient financing, technical difficulties, competition, difficulty scaling, and weak demand.

Even the giants themselves are cutting. OpenAI cut the video platform Sora—high operating costs plus low retention made it untenable; the browser Atlas was folded back into ChatGPT before surviving a full year; Operator and DALL-E were likewise folded into the main app, and even the "super app" redesign was rolled back amid user complaints.

Counting down, there is a long list of names: Relay, which built a Zapier alternative, was crushed by the platforms; Phind, a wrapper product, shut down just one month after closing its funding; Tome, which made presentations, had over 20 million users and $3 million in annual revenue; hardware pioneer Humane AI Pin raised $230 million and was ultimately bought by HP for $116 million; Yupp burned through $33 million; Figgs AI, free but too expensive to keep alive; top companion app Character.ai went bankrupt—the "graves" in the AI graveyard are still multiplying.

China is clearing the field as well: in 2025 alone, at least 25 domestic AI applications were shut down, delisted, or left without updates for long periods; according to AI Graveyard statistics, 392 AI tools worldwide stopped service that year. Tencent Yuanqi, QClaw, and the self-built agents of Doubao and Qwen were successively shut down and taken offline. They were merged with enterprise services into a unified "office" agent.

Autonomous driving has already died once

This list of autonomous-driving companies that exited from 2019 to 2026 is in fact an autopsy report on the first full cycle of the AI industry chain. Drive.ai, Starsky Robotics, Uber ATG, Argo AI (Ford and Volkswagen withdrew funding, the capital chain broke), lidar leader Velodyne (product involution), TuSimple (difficulty commercializing technology + geopolitical pressure + internal management, delisted), Luminar, Zongmu Technology (three failed IPOs), HoloMatic, QianGua Technology, Qingtian Smart Card... the causes of death are highly consistent: financing rupture, difficulty achieving commercial deployment, slow technology iteration, and tight capital chains.

These were not small workshops. Argo AI was backed by Ford and Volkswagen, Velodyne was once the industry leader, and TuSimple was the "first listed autonomous trucking stock." But the commercialization leg never landed, and once the capital market tightened, it snapped. In eight years, autonomous driving has proven one thing with a list of more than 40 exited companies: technology working does not equal business working.

Today's general-purpose large models are repeating autonomous driving's frenzy of 2019–2021, and the ebb tide that began in 2022—only the stage is bigger, the money burns faster, and the scale is more frightening.

Embodied intelligence, the next bubble?

The same script is already being rushed onto the stage. In August, DeepSeek took a 141 million yuan stake in Unitree Robotics, and Liang Wenfeng is set to lead another investment in the humanoid robot company Winverse AI. Embodied intelligence is standing exactly where autonomous driving once stood: hot financing, valuations running ahead, and commercialization nowhere in sight.

Robots burn more money than autonomous driving—hardware, supply chains, mass production, scenarios, every link is a money-burning black hole, and currently not a single company has a sustainable business model running. Worse, it shares the same lifeline as large models: capital. Once this round of the computing-power arms race cools first, embodied intelligence will be the first track to feel the chill of a supply cutoff. It does not even need to become a second autonomous-driving bubble itself—it only needs to walk the same path once more, and that is enough.

My judgment is no. They are the peak-seers of the dividend—mobile internet growth has peaked, new energy vehicles have been involutioned into overcapacity—and all are looking for the next certain growth engine. And AI, especially computing power, happens to be the certainty closest to "infrastructure" in the eyes of capital today. So what they are going all in on is not AI applications, but the AI foundation: one wants the traffic gateway, one wants electricity; in essence, both are staking out positions.

Risk and opportunity coexisting is a correct platitude, but concretely it comes down to two sides: the opportunity is that if DeepSeek can truly use hundreds of billions in capital to blast open a domestic computing-power ecosystem, with chips—energy—models—applications integrated as one, this will be an infrastructure transformation more foundational than the mobile internet; the risk is that if this round of investment, like autonomous driving, still only flowers and bears no fruit, the shock of a hundred-billion-level bubble bursting will far exceed the sum of those years in autonomous driving.

The endgame of the AI story will not be written in the Mag 7's financial reports, nor in financing news, but in the profit-margin tables of the S&P 493. Only when the economy truly catches the money AI is spending will it count. Before that, this hundred-billion all-in is more of a huge gamble on the future—win the bet and it is the foundation; lose it and it is a tombstone. The list in the AI graveyard is still getting longer. AI industry signal channel: Frontier Technology. This content is published with the author's authorization; the views represent only the author and do not represent Huxiu's position. If you have objections or complaints about this manuscript, please contact tougao@huxiu.com. Everyone who is changing and wants to change the world is on the Huxiu APP[ ]