May 25, 2026 · Money Circuit
ZOOM INVESTED $53 MILLION IN ANTHROPIC IN 2023. THREE YEARS LATER IT IS SITTING ON A $1.27 BILLION WINDFALL AND NOBODY SAW THIS COMING.
SOURCE: Bloomberg — May 22, 2026
Let me tell you about one of the quietest big wins in recent tech history. Zoom, the video call company that everyone used obsessively in 2020 and then sort of forgot about, placed a $53 million bet on Anthropic back in 2023. Not a big bet for a company their size. The kind of bet where you write the check, file the paperwork, and mostly stop thinking about it. A rounding error in your annual report. A line item that gets skimmed past in earnings calls.
This week, Bloomberg reported that Zoom disclosed in a regulatory filing that its Anthropic stake is now worth approximately $1.27 billion. They have netted about a billion dollars on the position. In three years. On a $53 million check.
That is a 23-times return. In three years. On a company that makes video conferencing software and had approximately zero strategic reason to be investing in AI labs in 2023 other than the fact that someone at Zoom apparently read the room correctly when almost nobody else did.
Here is what makes this story particularly enjoyable: Zoom spent most of 2022 and 2023 in a narrative of corporate decline. Post-pandemic hangover, shrinking user growth, losing the cultural cachet of the lockdown era. The stock had cratered from its pandemic highs. People were writing the “what went wrong with Zoom” articles. And somewhere in the middle of all that, Zoom quietly wrote a $53 million check to an AI startup that Dario Amodei and his colleagues had just founded after walking out of OpenAI.
The investment is still private equity on Zoom’s balance sheet, which means they have not cashed out. But the disclosed value gives you a real sense of what the position is actually worth right now, and the answer is: a lot. Analysts at Baird had previously estimated the stake could be worth $2 to $4 billion depending on Anthropic’s valuation trajectory. Given that Anthropic is now raising its latest round at a valuation above $900 billion, those estimates suddenly do not look crazy.
The reason this matters beyond the feel-good story of a struggling tech company getting lucky is what it tells you about where value has actually accumulated in the AI boom. Everyone talks about Nvidia’s stock price. Everyone talks about the startup funding rounds. But there are dozens of companies that made early bets on the right labs and are now sitting on positions that could redefine their balance sheets. Oracle has its OpenAI relationship. Amazon and Google both put billions into Anthropic. Zoom, of all companies, put in $53 million and ended up with a billion-dollar return.
The people who got rich in the California Gold Rush were not always the miners. Sometimes they were the guys selling shovels. And sometimes they were the guy who owned the plot of land next door to where someone else struck gold. Zoom did not build the AI. It just happened to be standing in approximately the right place at the right time with a checkbook. That is not a repeatable strategy. But it is a very good outcome.
The moral of the story, if there is one, is that in a technology transition this large and this fast, the benefits spread in ways that nobody fully anticipated. Zoom got a billion dollars richer because they believed in Anthropic before Anthropic had proven much of anything. In 2023 that looked like a minor strategic partnership. In 2026 it looks like one of the better investment decisions a mid-tier tech company has made in the past decade.
DEEPSEEK’S FOUNDER JUST TOLD INVESTORS THE CHATBOT WAS NEVER THE POINT. THE REAL GOAL IS AGI. AND HE IS RAISING TEN BILLION DOLLARS TO BUILD IT.
SOURCE: Bloomberg — May 22, 2026
So here is the thing about DeepSeek that everyone kind of suspected but nobody had confirmed until this week: the cheap, open-source AI models that rattled Silicon Valley earlier this year were never the endgame. They were the opener.
Bloomberg reported Friday that DeepSeek founder Liang Wenfeng told potential investors in the company’s ongoing 70 billion yuan funding round, which works out to roughly $10 billion, that the company’s actual mission is artificial general intelligence. Not building a better chatbot. Not winning the Chinese enterprise software market. AGI. The thing that Sam Altman at OpenAI has been saying he wants to build since 2015. The thing that Dario Amodei at Anthropic writes 15,000-word essays about. That thing.
Liang also told investors he intends to keep DeepSeek’s models open-source even as the company raises this enormous round, which is an unusual commitment for a company valued at what sources say is now approaching $45 to $50 billion. The lead investor in the round is China’s state chip fund, the same government vehicle that backs the country’s biggest semiconductor companies. So you have a hedge fund manager who built an AI lab as a side project, who is now raising $10 billion backed by the Chinese government, who says his real goal is to build artificial general intelligence, while publicly committing to give the research away for free.
That is a genuinely strange set of facts to hold together in your head.
The open-source commitment is not irrational. DeepSeek built its reputation entirely on releasing its models publicly, which is how it got the world’s attention in the first place. Closing that off now would be a betrayal of the community that made the company famous. And from a purely strategic perspective, having millions of developers worldwide building on your architecture is not a bad position to be in when you are trying to recruit the best researchers and attract the kind of visibility that gets governments and corporations to take you seriously.
The AGI declaration, though, is where things get interesting. Because if Liang is serious, and the evidence suggests he is, then this round is not about commercializing what DeepSeek has already built. It is about funding the next several years of frontier research with the explicit aim of reaching general AI. That puts DeepSeek in direct competition not just with Chinese labs but with OpenAI, Anthropic, and Google DeepMind on the most consequential race in the history of technology.
The geopolitical dimension of that is something the Pentagon and the State Department are presumably already thinking about. A state-backed Chinese lab that is also open-source and also explicitly targeting AGI is a fairly unusual combination of properties. The open-source part reassures some people and terrifies others. The AGI ambition part mostly just terrifies people.
What Liang seems to understand is that AGI is not a feature you ship in a product update. It requires enormous sustained investment in compute, talent, and research infrastructure over many years. The $10 billion round is presumably meant to buy that runway. Whether it is enough, and whether DeepSeek can attract the talent and compute needed to actually compete at the frontier, are different questions. But the ambition has now been stated plainly, in meetings with actual investors, with actual money on the table. This is not a research project anymore. It is a race.
MARKET STRATEGISTS ARE DRAWING DOT-COM PARALLELS AS SPACEX AND OPENAI PREPARE TO FLOOD WALL STREET WITH THE LARGEST IPOS IN HISTORY. “IF THEY CANNOT MAKE MONEY, THE WHOLE THING FALLS APART.”
SOURCE: CNBC — May 22, 2026
Everyone at the party is having a great time. The music is loud, the drinks are flowing, and two of the most anticipated IPOs in the history of American finance are about to arrive at the door. SpaceX is targeting June 12. OpenAI is aiming for September. Between them, they will be asking public market investors to absorb something close to three trillion dollars in combined valuation. And a handful of market strategists are starting to quietly wonder whether this is the moment to find your coat.
CNBC reported this week that a cluster of portfolio managers and analysts are now drawing explicit parallels between the current AI IPO wave and the dot-com era of the late 1990s. William de Gale at BlueBox Asset Management put it plainly: “If OpenAI and Anthropic cannot make money, this whole thing falls apart.” Jim Cramer, who is never accused of subtlety, called SpaceX’s upcoming debut potentially “destructive” for the rest of the market and warned it could become “a bubble unto its own.”
The specific concern being raised by the skeptics is what some are calling a “supply event.” When SpaceX lists at $1.75 trillion and OpenAI shows up a few months later at $850 billion, institutional portfolio managers are going to have to find a very large amount of money to participate in those floats. That money has to come from somewhere. The most likely somewhere is: the rest of the stock market. Funds rebalance. Other positions get sold down to make room for the new arrivals. The sheer gravitational pull of two trillion-dollar listings in the same calendar year could create turbulence for everything else in the portfolio.
There is also the earnings question, which is a lot less exciting to discuss but is probably more important. SpaceX’s IPO filing disclosed that the company’s AI division absorbed xAI and burned more than six billion dollars in 2025 on about three billion in revenue. OpenAI is generating serious revenue now but has not posted a quarterly profit yet. The bet that public market investors are being asked to make is that these companies will grow their way into their valuations fast enough to justify the multiples. That is a bet the market has been willing to make before. Sometimes it works out. Sometimes you end up with a lot of Pets.com shares.
None of this means the AI bubble is about to pop. The companies are real, the revenue is real, and the technology is genuinely transformative in ways that the dot-com companies mostly were not. But there is a difference between “this technology will change everything” and “these specific companies at these specific valuations are a good buy on their first day of trading.” History is full of transformative technologies whose early investors got crushed anyway because they paid too much too soon.
The optimists, who still outnumber the skeptics by a wide margin, will tell you that this time is different because the revenue is already there. OpenAI is generating over $40 billion annualized. Anthropic is posting quarterly operating profits. These are not vaporware companies. That is fair. But “different this time” is also the most dangerous phrase in finance, and the people saying it most loudly tend to be the ones with the most to lose if they are wrong. Worth watching. Very worth watching.
FORMER CITADEL QUANTS JUST RAISED $78 MILLION TO BUILD AN AI THAT READS THE MARKETS THE WAY THE BEST TRADERS DO. ANDREESSEN HOROWITZ AND INDEX VENTURES BOTH WROTE CHECKS.
SOURCE: Bloomberg — May 19, 2026
Citadel is one of the most successful quantitative hedge funds in the world. Ken Griffin built it into a machine that generates returns by hiring the best mathematicians and coders on earth, giving them access to better data and faster infrastructure than their competitors, and then running systematic strategies at a scale and speed that human traders cannot match. Getting hired at Citadel as a quant is hard. Getting hired there and staying long enough to be considered senior is harder. Leaving with the kind of reputation that lets you raise $78 million for your own startup is something that happens a small number of times per decade.
Bloomberg reported Monday that a group of former Citadel quants did exactly that. Their company is called Moment, and it raised $78 million in a round led by Index Ventures, with Andreessen Horowitz also participating. The pitch, at the highest level, is that modern AI can process and synthesize financial information in ways that give it an analytical edge in markets, and that the people best positioned to build and deploy that AI are people who have spent careers inside the most sophisticated trading operations in the world.
This is a different category of AI startup than the ones that dominate most of the coverage. Moment is not building a general-purpose language model. It is not competing with OpenAI or Anthropic for enterprise contracts. It is building something much more specific: AI infrastructure for financial decision-making. The people who built it understand financial markets from the inside in a way that most AI developers simply do not.
The reason this story matters for Money Circuit specifically is what it signals about where AI is going in finance. For years, the conversation about AI in financial services was mostly about chatbots for customer service, fraud detection in banking, and automation of back-office processes. Those are real applications, but they are not where the serious money is made in finance. The serious money is made in trading, in risk management, in identifying mispricings before other sophisticated participants do.
That is the territory Moment is entering. And the fact that Index Ventures and a16z both backed the round suggests the venture community believes there is a genuinely large opportunity here. a16z in particular has been careful about fintech investments in recent years, so a participation check in a round like this is a meaningful signal.
The interesting question is whether AI can actually do what the best quants do, or whether it can merely approximate it. Citadel’s edge has never been just raw computing power. It has been the combination of computing power, proprietary data, and the judgment of people who understand market microstructure at a deep level. Moment is betting that its founders can encode enough of that judgment into their systems to build something durable. They have $78 million and some very good backers to find out if they are right. The quants are betting on themselves. They tend to be pretty good at calculating odds.
A 25-YEAR VENTURE CAPITAL VETERAN IS NOW PUBLICLY SAYING HIS INDUSTRY LOCKED ORDINARY AMERICANS OUT OF THE BIGGEST WEALTH CREATION MOMENT IN TECH HISTORY. AND HE SAYS IT WAS NOT AN ACCIDENT.
SOURCE: Fortune — May 22, 2026
OpenAI is currently valued at $852 billion. Anthropic is raising at $900 billion. SpaceX is targeting a $1.75 trillion listing. Between them, roughly four trillion dollars of value has been created, and almost none of it has been accessible to ordinary retail investors. The gains have gone to venture capitalists, sovereign wealth funds, institutional allocators, and a small number of accredited investors with access to private market vehicles. The person who drives for Uber, the teacher in Phoenix, the nurse in Cleveland who has a 401(k) and watches the AI news and thinks they should be part of this, has been almost entirely excluded from the wealth creation that has happened since 2022.
A veteran of 25 years in venture capital wrote about this in Fortune this week, and what makes the piece notable is that it is an insider saying plainly what outsiders have suspected: this is not an accident of market structure. It is the product of deliberate choices made by an industry that benefits from keeping private markets private. The longer a valuable company stays private, the longer the returns stay concentrated among the handful of funds and individuals who have access to private equity and venture capital.
The numbers are staggering when you look at them in aggregate. In Q1 2026, AI companies absorbed roughly $242 billion in venture funding, representing about 80% of all global venture capital deployed in the quarter. Four of the five largest venture rounds in history closed in that single quarter alone. The wealth creation is happening at a scale and speed that has no precedent in the history of private markets. And the person with a brokerage account at Fidelity cannot touch any of it.
There are structural reasons for this that are not entirely nefarious. Private companies are not required to make the same disclosures as public ones. There are accreditation rules that theoretically protect unsophisticated investors from high-risk bets. Venture capital is a long-duration, illiquid asset class that genuinely does not work well in the same format as a retail stock. These are real considerations.
But they also happen to serve the interests of the people who profit most from private markets staying exclusive. And as the piece argues, there are alternative structures, sovereign wealth fund models, interval funds, retail venture vehicles, that could expand access without the regulatory fiction that someone making $250,000 a year is inherently capable of bearing risks that someone making $80,000 cannot. The accredited investor threshold is a blunt instrument that correlates wealth with sophistication in a way that is more convenient than accurate.
The irony is that as OpenAI and SpaceX approach their public listings, retail investors will finally get their shot. But they will be buying in after the venture funds have already made 100x. They will get the last act of the play, not the first. The curtain went up in 2020, 2021, 2022. The insiders were in the room. Most people were not. That is the story of every major technology cycle in American history, and based on the evidence, there is no particular reason to think this one is going to end differently.
Unless someone actually changes the rules. Which, historically, is not the most reliable outcome to bet on.
Keywords: AI investment 2026, Zoom Anthropic stake billion, DeepSeek AGI goal $10 billion round, AI IPO market top warning dot-com bubble, Citadel quants AI fintech Moment, venture capital wealth inequality AI gold rush, Money Circuit weekly AI finance