NVIDIA JUST BROKE EVERY RECORD IN THE BOOK: $81 BILLION IN ONE QUARTER, AN $80 BILLION BUYBACK, AND JENSEN HUANG RAISED THE DIVIDEND FROM A PENNY TO 25 CENTS

SOURCE: Fortune — May 20, 2026 Let me give you a number and then we can talk about it. Eighty-one point six billion dollars. That is what Nvidia generated in revenue during its most recent quarter, the three months ending April 2026. That is more money than the entire annual GDP of several medium-sized countries. It is more revenue than every NBA team combined makes in a full year. It is an 85 percent increase from the same quarter last year, which was itself a record, which was itself a record before that. Jensen Huang has been printing money so consistently that it is starting to feel like a natural phenomenon, like a river that refuses to stop flowing. The numbers are genuinely hard to process. Analysts had estimated roughly $79 billion in revenue. Nvidia did $81.6 billion. Analysts expected earnings of $1.77 per share. Nvidia delivered $1.87. The data center business, which is the AI chip business, grew 73 percent year over year. The company said it has more orders than it can fill. More orders than it can fill. A company generating $81 billion in quarterly revenue is still supply-constrained. The demand for what Nvidia makes is so large that they physically cannot build chips fast enough to give them to all the customers who want them. And then Huang announced a buyback. Not a small one. Eighty billion dollars. The company is going to buy back $80 billion of its own stock, which is a statement that says two things simultaneously: first, we believe our stock is undervalued relative to our future earnings, and second, we have so much cash that we can spend a figure that would be the third or fourth largest acquisition in corporate history just on buying our own shares. The dividend was raised from one cent per share to twenty-five cents per share, which is a 2,400 percent increase. There are people on Wall Street who spent 2023 and 2024 writing theses about how Nvidia’s growth would slow when the hyperscalers finished their buildout. Those people have been wrong for eight consecutive quarters. The AI infrastructure buildout is not slowing. Microsoft is targeting $190 billion in capital expenditure this year. Google is at $185 billion. Amazon, Meta, and every other major tech company is spending more, not less. All of that spending flows, at some point, through Nvidia’s chips. The TAM is not shrinking. If anything, it is getting larger as AI moves from training runs done by a handful of labs to inference deployed at consumer scale across every major enterprise. The stock has done something remarkable over the past three years. It has gone from a gaming and data center chip company that smart investors liked to the single most important piece of infrastructure in the most consequential technology transition in a generation. Jensen Huang, who is a relentlessly cheerful man who wears the same leather jacket to every event, has guided a company from interesting to indispensable. When the history of this era gets written, Nvidia will have a chapter. That chapter will contain a lot of very large numbers. The only honest question is whether the growth can continue, and the honest answer is that so far every quarter someone asks that question, the answer is yes. The company said revenue will be approximately $87.5 billion next quarter. That is another record. —

SOFTBANK ADDED $61 BILLION TO ITS MARKET CAP IN TWO DAYS AND ALL MASAYOSHI SON HAD TO DO WAS OWN OPENAI BEFORE EVERYBODY ELSE DID

SOURCE: CNBC — May 22, 2026 | Bloomberg — May 25, 2026 If you had told Masayoshi Son in 2022 that his company would add $61 billion to its market cap in two trading days, he would have laughed and then immediately asked if it could be more. Son has been making very large bets for a very long time, some of which have gone terribly (WeWork, the $47 billion disaster), and some of which have gone so well that they retroactively make you forget the disasters. His bet on OpenAI is in the second category. Here is what happened. On May 21 and 22, SoftBank shares in Tokyo surged by almost twenty percent over two trading sessions, adding over $61 billion in market cap, and then continued to climb to record highs by May 25. The catalyst was confirmation that OpenAI is moving toward an IPO and that SoftBank Energy is also preparing to list in the US. Two companies that SoftBank is deeply invested in are about to go public, and the market looked at SoftBank’s holdings and did some math. The math is compelling. SoftBank has put more than $30 billion into OpenAI since 2023. Their Vision Fund reported gains of $46 billion from the OpenAI stake in the fiscal year ending March 2026. That is not a typo. Forty-six billion dollars in paper gains from a single investment. The Vision Fund, which was famous for losing money on half the companies it funded at preposterous valuations, found its redemption arc in a nonprofit-turned-for-profit AI lab that Sam Altman built into the most talked-about company in the world. The two-day surge reflects something important about how the market is pricing the AI gold rush. SoftBank is not an AI company. It makes no chips, trains no models, writes no code. It is a Japanese holding company that owns stakes in technology companies and has a complicated balance sheet. But because it owns a large piece of OpenAI, it has become a way for global investors to get exposure to the AI trade through a publicly traded vehicle. When OpenAI’s IPO moves closer, SoftBank’s valuation moves up. When OpenAI raises at higher valuations, SoftBank looks smarter. Masayoshi Son has said publicly, and with his characteristic absence of understatement, that artificial general intelligence will arrive soon and that it will be the most transformative event in human history. He said this after the WeWork collapse, after the SoftBank Vision Fund had written down billions in investments, after the skeptics had written many very confident articles about the end of the SoftBank era. He kept making AI bets. He put $30 billion into OpenAI when most institutional investors were watching from the sidelines, trying to figure out whether the hype was real. The hype turned out to be very, very real. And now SoftBank is adding $61 billion in two days while the record-high shares sit in Tokyo and Son prepares for the IPO that will, if it goes as planned, produce one of the largest single investment payoffs in the history of financial markets. The WeWork money is still gone. But when you are sitting on $46 billion in OpenAI gains, you stop talking about WeWork. —

SNOWFLAKE STOCK SOARED 36 PERCENT IN ONE DAY AFTER SIGNING A $6 BILLION DEAL WITH AMAZON AND SUDDENLY THE MARKET REMEMBERED THAT SOFTWARE COMPANIES STILL EXIST

SOURCE: CNBC — May 27, 2026 | TechCrunch — May 27, 2026 For about six months, the working assumption among technology investors was that AI would kill enterprise software. Not immediately, but eventually, and maybe sooner than anyone wanted to admit. The logic was clean: if a language model can do the analytical and workflow tasks that software platforms charge for, why would you keep paying for the software? Snowflake, which is a cloud data platform that charges companies to store, organize, and query their data, was getting hit by this thesis. The stock was down significantly from its peaks. The narrative was, politely, not great. And then on Wednesday Snowflake announced that it has signed a five-year, $6 billion commitment with Amazon Web Services. The company will migrate a significant chunk of its infrastructure to Amazon’s Graviton chips, the custom Arm-based processors that Amazon has been developing for years as an alternative to standard hardware. Snowflake also reported that it expects product revenue to grow about 31 percent in the fiscal year ending January 2027, beating analyst estimates. The stock went up 36 percent that day. Its best single-day performance ever. And then it kept pulling the rest of the software sector with it. ServiceNow, Oracle, and Palantir all gained more than six percent the same day. Microsoft added three percent. Atlassian joined the party. The entire software category, which had been sitting in a quiet existential crisis about whether AI was a friend or an enemy to their business model, collectively got a little bit of its confidence back. Here is what the Snowflake deal actually says about the AI moment. The concern was that AI replaces SaaS. What the deal shows is that AI needs SaaS to scale. Snowflake’s entire value proposition is organizing enterprise data in a way that AI models can actually use. If you want to build AI agents that understand your company’s sales history, customer behavior, supply chain, and financial data, you need that data to be clean, structured, and accessible. That is exactly what Snowflake does. The company is not being disrupted by AI. It is becoming the data infrastructure on which enterprise AI runs. The Amazon relationship matters on multiple levels. Snowflake committed to using Amazon’s Graviton CPUs, which is a bet that custom silicon from cloud providers is going to become a serious alternative to the GPU-dominated AI compute stack. Amazon has been signing up major customers for Graviton, including Meta, which announced deals for millions of Graviton chips in April. Snowflake is now another major endorsement for the strategy. This is not just a cloud spending deal. It is a signal about where the enterprise AI infrastructure market is heading. For the people who had written off Snowflake, the 36 percent day was uncomfortable. It is always uncomfortable when a company you wrote a bear thesis on announces that it has figured out the exact thing you said it could not figure out. But the broader lesson is worth noting: the assumption that AI destroys rather than transforms enterprise software was, at minimum, premature. The enterprise needs infrastructure for its AI ambitions. The companies that provide that infrastructure are not being replaced. They are being upgraded. —

A TOP PRIVATE EQUITY INVESTOR TRAINED AN AI ON 13 YEARS OF DEALS HIS FIRM HAS DONE AND NOW THE THING ARGUES BACK WHEN IT THINKS THE NUMBERS ARE WRONG

SOURCE: Fortune — May 26, 2026 There is a very particular type of knowledge that exists inside a private equity firm and almost nowhere else. It is the accumulated judgment of dozens of senior investors who have looked at thousands of companies over decades, made bets on which ones to buy and which ones to pass on, and then lived with the consequences of those bets. The institutional memory of a firm like Advent International, which manages roughly $100 billion and has been doing complex buyouts since the late 1980s, is not written down anywhere complete. It lives partly in spreadsheets and investment memos, but mostly in the heads of people who have been doing this for a long time. James Brocklebank, who co-chairs Advent, decided to change that. The firm spent years organizing its investment committee papers, every memo written about every deal considered over 13 years, into a format that an AI could be trained on. Not just the deals they did. Also the deals they passed on. The reasons they said no. The assumptions that were wrong. The margins they projected that did not hold. The market dynamics they misread. The result is something Advent calls the IC Robot. When a new deal comes before the investment committee today, the IC Robot reads the memo and runs it against everything the firm has ever argued about. If a manager projects that a healthcare services company will expand its operating margins from 15 to 22 percent over four years, the IC Robot checks that assumption against every similar bet Advent has made historically. It can say, in effect: this type of company, in this type of market, with these operating characteristics, has hit margin targets like this only about 40 percent of the time. Here are the three that did it and here are the five that did not and here is what was different about each one. Brocklebank described it on Goldman Sachs’s podcast this way: the IC Robot can say things that a junior analyst does not feel comfortable saying to a senior partner. It has no career to protect. It has no relationship with the deal team that spent six months on the memo. It just has 13 years of outcomes. And it will tell you, politely but without hesitation, when your assumption does not match the historical pattern. This is the application of AI to finance that gets very little coverage because it is not a funding round and it is not a valuation. But it is arguably more significant than most of the things that do get covered. The firms that get AI working as a genuine analytical layer inside their investment process, not as a marketing talking point but as an actual check on human judgment, are going to have an informational advantage over the firms that are still running purely on senior partner intuition. Brocklebank said something that deserves to sit with you for a minute. He said his firm is moving toward a world where they are working for the AI rather than the AI working for them. Not because the AI is taking over, but because the AI’s effectiveness depends on how well you organize your data and your processes around it. If you want the IC Robot to work well, you have to feed it well, structure your memos consistently, track your assumptions rigorously, and document your outcomes honestly. The AI makes you more disciplined. It makes the whole firm more disciplined. That is the part of this story that does not make headlines but is quietly changing how serious money gets managed. —

UBER BURNED THROUGH ITS ENTIRE 2026 AI CODING BUDGET IN FOUR MONTHS AND ITS OWN COO WENT ON A PODCAST AND SAID HE CANNOT TELL IF ANY OF IT WAS WORTH IT

SOURCE: Fortune — May 26, 2026 There are two kinds of AI stories being told right now. The first kind is the kind you read about every week in Money Circuit: the billion-dollar raises, the record valuations, the quantum leaps in model capability. The second kind of story is harder to find but more honest: what actually happens when a big company tries to use all that AI in the real world. Uber ran into the second kind of story in May. Uber is about as AI-forward as a Fortune 500 company can be. The entire business is algorithmic: surge pricing is AI, route optimization is AI, fraud detection is AI, driver dispatching is AI. These are not bolt-on features. They are the core product. So when Uber started rolling out AI coding tools to its engineering teams, the company had every reason to expect strong results. They created an internal leaderboard tracking which teams used AI tools the most. They encouraged adoption. They made it visible and competitive. Engineers responded. By April, Uber had burned through its entire 2026 AI coding budget. Four months into the year, the whole budget was gone. That is when Andrew Macdonald, the company’s president and COO, went on a podcast and said something that most executives in his position would never say out loud. He said it is very hard to draw a line between the AI spending and useful features being shipped to customers. He said, specifically: “That link is not there yet.” The inference is clear: Uber spent the whole year’s coding AI budget in four months, and the COO cannot tell you what they got for it. This is not a story about AI failing. It is a story about the gap between capability and measurable output. Uber’s engineers are using AI coding tools. Code is being written faster. Tests are being run. Commits are being merged. But somewhere between all that activity and the customer opening the Uber app and having a better experience, the accountability chain breaks down. The tools are working. The ROI is unclear. Microsoft ran into a similar problem on the same week. The Verge reported that Microsoft had begun canceling most of its direct Claude Code licenses, moving engineers toward GitHub Copilot CLI instead. Uber, Microsoft: two companies with enormous technical sophistication, enormous financial resources, and a genuine commitment to AI adoption, both hitting the same wall. How do you justify the spend when you cannot measure the output? This is the most important question in enterprise AI right now and almost nobody in the financial media is asking it, because the financial media is distracted by the funding rounds. But the companies buying the $200 billion in AI tools that Gartner predicts will be sold this year are going to start demanding answers. You can spend through your budget in four months on tools that make your engineers feel productive. But at some point the CFO wants to see the number. The COO wants to draw the line. And right now, in a lot of companies, that line is not there yet. The AI hype is real. The technology is real. The productivity gains in certain contexts are also real. But the current enterprise AI boom contains a very large amount of money being spent by companies that are not yet certain they are getting value. Some will figure it out. Others will quietly stop renewing the contracts. The reckoning is coming, and when it does, it will be a much more interesting story than another funding round. —

THE WOMEN RUNNING THE MONEY AT META, MICROSOFT, GOOGLE, OPENAI, ORACLE, AND NVIDIA ARE AUTHORIZING MORE CAPITAL SPENDING THAN MOST COUNTRIES WILL EVER SEE AND NOBODY MENTIONS THEM BY NAME

SOURCE: Fortune — May 27, 2026 Here is a fact that is simultaneously obvious and completely overlooked. The people signing off on the largest corporate capital expenditures in the history of technology are almost all women. Meta’s CFO is Susan Li. Microsoft’s CFO is Amy Hood. Alphabet’s CFO is Anat Ashkenazi. Oracle’s CFO is Hilary Maxson. OpenAI’s CFO is Sarah Friar. Nvidia’s CFO is Colette Kress. Between them, they are overseeing somewhere north of $700 billion in capital spending this year alone. That is not a rounding error. That is more than the GDP of most countries. Fortune published a piece this week noting this fact and exploring what it means, and the answers the executives gave were characteristically varied. Some see it as a milestone. Some describe it as a coincidence. Some decline to read too much into it. Colette Kress at Nvidia, who has been CFO since 2013 and has overseen the company’s transition from a gaming chip maker to the infrastructure backbone of the AI era, has watched the company’s market cap go from around $8 billion when she joined to roughly $3.3 trillion today. Her signature has been on every quarterly report, every earnings call, every major financial decision during one of the most extraordinary corporate growth stories in modern history. Amy Hood at Microsoft is navigating a $190 billion capital expenditure plan for 2026, a 61 percent jump from the prior year. To put that in context, Microsoft is spending more on infrastructure this year than the entire US interstate highway system cost to build, adjusted for inflation. Hood has been CFO since 2013 and has been one of the quiet architects of Microsoft’s transformation from an enterprise software company into a cloud and AI giant. She does not get the press coverage that Satya Nadella does, but she is the one ensuring the numbers work. Susan Li at Meta joined the CFO role in 2022, right before Mark Zuckerberg announced the “year of efficiency” that involved cutting 25,000 jobs. She managed the brutal cost discipline of 2022 and 2023 and is now overseeing the $145 billion capital expenditure plan for 2026, which represents Zuckerberg’s full reversal toward aggressive AI investment. Sarah Friar at OpenAI is the newest to the list, having joined as the company transformed from nonprofit lab to the most anticipated IPO in American history. The Fortune piece notes that these women are cutting against what researchers call the “glass cliff,” the pattern where women get handed leadership positions primarily in moments of crisis. These CFOs arrived at moments of scale and ambition, not collapse. They were brought in, or kept in place, during periods when the companies were at or approaching their peaks. The irony is that while the AI industry generates enormous coverage of the CEOs, the founders, and the investors, the people actually authorizing the capital decisions that are reshaping the global economy are largely invisible to public discourse. Jensen Huang is on the cover of magazines. Colette Kress is signing the documents that make the numbers in Jensen Huang’s presentations real. Sam Altman is giving keynote speeches about the future of intelligence. Sarah Friar is structuring the IPO that will take that vision to the public markets. The money and the people who manage it are the actual machinery of the AI revolution. It just does not photograph as well as a leather jacket. Keywords: Nvidia Q1 2026 earnings record revenue AI chips buyback, SoftBank OpenAI IPO market cap surge, Snowflake AWS Graviton AI deal stock surge, Advent International IC Robot private equity AI, Uber AI budget spent four months ROI, women CFOs Big Tech AI spending Meta Microsoft Google Nvidia OpenAI Oracle, Money Circuit weekly AI finance