Peter Thiel is betting that Cognition's three-month valuation has skyrocketed by more than 50%! The main line of AI investment spreads from “shovel sellers” to “who uses shovels to extract profits”
The Zhitong Finance App learned that some media quoted information revealed by people familiar with the matter as reporting that Cognition AI Inc., a leader in AI programming intelligence supported by Peter Thiel, the godfather of venture capital in Silicon Valley, is in preliminary negotiations with investors on a new round of financing. This round of financing may increase the valuation of this AI programming startup by at least 50% to at least 40 billion US dollars.
The main investment line in the stock market is currently shifting from “who allocates and builds the largest GPU data center” to “who can turn tokens into sustainable cash flow”. This is the underlying logic that Goldman Sachs is optimistic about Microsoft becoming the core beneficiary of the “AI application monetization stage.”
The super-bull market surrounding AI is gradually moving from “buying chip stocks” to “buying AI workflows”, that is, the current market is repricing the main AI investment line from “who invests the most capital expenditure” to “who can convert computing power to ARR, profit margin, and free cash flow the fastest”. This latest rotation is conducive to software companies that embed key enterprise processes and focus on AI application platforms with high renewal rates, data barriers, and intelligent monetization capabilities.
Recently, there has been a very clear trend of “single-line AI-themed transactions from AI computing power infrastructure to monetization and diffusion” within US technology stocks, but it cannot be said that capital has completely withdrawn from AI computing power themes and semiconductors and then switched to AI application software. The iShares Software ETF (ETF code: IGV), which is a global leader in AI application software companies such as Microsoft, Google, Palantir, and Saftex, can be called the global “AI application investment trendsetter”. The ETF bucked the trend and rose 4.4% in July when global technology stocks experienced a pullback. During the same period, SOXX Semiconductor ETF (Philadelphia Semiconductor ETF) plummeted 22.1%, and NASDAQ 100 fell 6.6%, highlighting that the beta of the AI application software sector began to rise again.
The valuation of the AI programming circuit is booming again! Cognition hits $40 billion, Devin ignites “Smart Body Engineer” capital competition
According to information, people familiar with the matter said that less than three months after the AI startup intelligent programming company completed $1 billion financing at a valuation of 26 billion US dollars, potential Wall Street institutional investors have already begun contacts around the company. Since the relevant transaction is still private information, people familiar with the matter requested anonymity. Potential investors are seeking a share in one of the world's leading companies in one of the hottest technology sectors. According to people familiar with the matter, an important indicator for measuring the company's revenue generation related to AI applications — that is, the company's current annualized revenue operating rate index is close to 1 billion US dollars, which is about double the level at the time of the previous round of financing.
One of the people familiar with the matter said that Cognition is expected to raise more than 1 billion US dollars in this round of financing. People familiar with the matter said that the company may still decide not to finance or seek financing under different terms. A Cognition representative declined to comment.
Cognition sells an AI intelligent programming system called Devin, which can help enterprises or individual users automate a series of extremely complex engineering tasks. The company was founded in 2023 and has received support from top venture capital institutions such as General Catalyst and Founders Fund under Peter Thiel, the godfather of Silicon Valley venture capital.
Using AI agents for full-process proxy workflows and fully automated programming has probably become the most popular racetrack in the field of artificial intelligence applications. The strongest leaders in AI applications such as Anthropic PBC, OpenAI, and SpaceX have invested a large part of their business into their own project-level engineering products. SpaceX has agreed to acquire AI programming startup Cursor this year, and the deal is expected to close as soon as this week.
If Cognition AI chooses a new round of financing at a valuation of $40 billion or more, the AI startup will officially surpass the strong valuation Cursor received in the private equity market prior to the SpaceX deal. Cursor's previous valuation was around US$29.3 billion, but after achieving annualized revenue of over US$2 billion a month ago, it negotiated financing at a valuation of 50 billion US dollars in March.
Cognition AI itself is one of the purest targets in this “Agentic Software (AI agent-led software)” investment line. Its core product, Devin, is positioned as an autonomous AI agent software engineer who can independently plan, write, test, and deliver production code in a real code base and development tool environment.
According to information, after completing the acquisition of Windsurf, the company also formed a complete AI application software engineering platform “Windsurf/Devin AI agent collaborative development+Devin responsible for asynchronous autonomous execution in the cloud+Devin for Terminal undertaking complex engineering tasks”; Devin is a “cloud autonomous AI engineer”, Devin for Terminal is a “local command line entry”, and Devin Desktop is a desktop IDE where managers collaborate with multiple agents/ Console; the three are used jointly by customers and engineering teams. For example, Mercedes-Benz (Mercedes-Benz) has deployed its products to global R&D and IT systems. In a four-week pilot, Devin analyzed more than 200,000 lines of COBOL code, shortening the modernization work originally anticipated in eight months to eight days.
From “selling computing power” to “selling productivity”! The AI application layer takes over infrastructure and becomes the new main battleground for capital
From an investment perspective, the real importance of Cognition AI's valuation soaring to $40 billion in AI investment lines after only three months is that the capital market is beginning to directly price “how much human software engineering hours can be replaced by AI intelligence and how much measurable productivity” — if this trend continues, the AI application layer is likely to become the next battle for the largest profit pool in the entire AI supercycle, following the frenzy of GPU demand and the frenzy of data center infrastructure.
Cognition's AI valuation surged from $26 billion to at least $40 billion in less than three months, a very strong sign that “the AI investment boom is rapidly spreading from basic models and computing power infrastructure to the AI application layer (application layer)”, but more accurately, it is a revaluation of AI applications driven by real revenue growth and increasingly emphasizes the ability to deliver on ROI, rather than an indiscriminate bubble of all software companies.
Looking more broadly, the AI application layer is simultaneously crossing the three critical points of “user size, depth of use, and commercialization.” According to the latest official data from OpenAI, about 1 billion people use ChatGPT every week; US tech giant Google confirmed on August 11 that the Gemini App had surpassed 1 billion monthly users and became the fastest growing product in Google history. However, it is important to note that the two are weekly active and the other is monthly active, and the statistical caliber cannot be directly compared horizontally. At the same time, Anthropic is showing a more important enterprise-side monetization curve: recent reports show that its annualized revenue run rate (RRR) has exceeded 47 billion US dollars, and the AI intelligent programming tool Claude Code is an important driving force in this. RRR is the annualization of revenue from the most recent period. It is not equal to the revenue already realized for the past 12 months.
These three growth curves that have reached an important tipping point show that generative AI/AI agents are moving from a “user trial tool” to the high-frequency consumption+enterprise workflow embedding+agent execution task stage; from an AI system engineering perspective, this is why Coding Agent (AI agent programming application) has taken the lead — code is naturally digitized, output can be tested in a short period of time, ROI can be quantified, and the feedback loop is extremely short, so it is likely to become one of the earliest vertical scenarios in the Agentic AI field to achieve large-scale commercial closure.
The stock market has also begun to reward application software that can prove the ability of AI to redeem revenue, rather than generally rewarding the “software+AI” label. IGV, a software ETF, had a strong contrarian rise in July, and Palantir's latest quarterly revenue growth of 93% year over year to US$1.94 billion, and raised its annual revenue guide again. Demand for AI applications in the US continued to expand rapidly. After the financial report, the stock price rose 14% after the market and then rose close to 40% that week; this shows that once cutting-edge AI technology can be directly converted into contracts, customer expansion, and cash flow, the market is willing to give a very high growth premium.
If you look further at the round of rebound from July 31 to August 11, IGV rose from about $94.58 to $103.92, with a cumulative increase of about 9.9%; during the same period, SOXX (Philadelphia Semiconductor Index ETF) rose from about 527.10 to around $534, only +1.3%, and QQQ (NASDAQ 100 ETF) rose from 687.99 to around $718.45, about +4.4%. IGV achieved excess profits of about 8.5 percent and 5.4 percent, respectively, highlighting that the AI bull market gradually went “wild buying” “Chips” are moving towards “AI application monetization”.
The super-bull market surrounding AI is gradually moving from “buying chip stocks” to “buying AI workflows”. That is, the market is currently repricing one of the main tenets of AI bull market investment from “who invests the most AI capital expenses” to “who can quickly convert computing power into ARR or RRR, profit margin, and free cash flow”. This latest rotation is conducive to software companies focusing on AI application platforms that embed key enterprise processes, have high renewal rates, data barriers, and intelligent monetization capabilities, but this does not mean that all traditional software stocks will rise at the same time.
Global funding is spreading from GPU/HBM/data centers in the first phase of AI infrastructure to second-stage application winners that can turn tokens into enterprise productivity, revenue, and cash flow. In the future, valuation differentiation is likely to be even more intense: software companies with exclusive data, workflow entry, closed loop execution of agent agent workflows, and clear ROI will be re-evaluated, while traditional SaaS, which is prone to commercialization of basic model functions, may continue to be pressured. It is worth noting that the semiconductor sector remains the absolute winner throughout the year, but the strong performance of cloud computing giants and AI application leaders such as Palantir drove an accelerated shift in the global capital trend from “AI Capex growth” to “AI Capex return on investment”. Therefore, if this round of IGV's excess earnings compared to SOXX and QQQ continues to expand, then it will be a very important market signal: the AI supercycle is officially entering the second investment phase of “who uses a shovel to extract profits”.