MONEY CIRCUIT: GOOGLE SELLS STOCK LIKE A STARTUP, DEEPSEEK CASHES IN, AND ANTHROPIC IS WORTH ALMOST ONE TRILLION DOLLARS

ALPHABET RAISES $85 BILLION IN BIGGEST EQUITY SALE IN CORPORATE HISTORY TO FUND AI SPENDING BINGE

Let me put this in perspective for you. The previous record for the largest equity raise in history was held by Petroleo Brasileiro, the Brazilian oil company, which pulled in $70 billion back in 2010. That record stood for 16 years. Google’s parent company just smashed it in a single week.

Alphabet announced on June 1st that it was selling $80 billion worth of stock to fund its artificial intelligence infrastructure buildout. Then, two days later, after the market showed up and wanted more, the company upsized the whole thing to $85 billion. The final tally landed at $84.75 billion. They could not even wait a full 72 hours before going back to the well for an extra five billion.

Here is how the deal is structured. There is a $30 billion chunk in underwritten public offerings, which includes some mandatory convertible preferred stock. Then there is a $10 billion private placement going straight to Berkshire Hathaway. And then there is a $40 billion at-the-market program, which means Google will just be quietly selling shares into the open market for the rest of the year like some kind of perpetual fundraising machine.

The Berkshire angle is the really interesting part. Warren Buffett spent decades dismissing tech companies as uninvestable because he claimed he could not understand them. His successor, Greg Abel, just wrote a $10 billion check to Google in what is essentially his first major capital allocation decision as the person running Berkshire’s money. Berkshire has been sitting on nearly $400 billion in cash. Parking $10 billion of it in a company that owns the world’s dominant search engine, a growing cloud business, and a credible shot at AI leadership is not exactly a crazy bet.

Alphabet says it needs the money for capital expenditures. The company revised its 2026 capex forecast to between $180 billion and $190 billion, up from an already eye-watering $175 to $185 billion estimate made earlier in the year. They are building data centers and buying chips at a pace that would make a defense contractor blush. The thinking is simple. If AI is going to be the next computing platform, the company that owns the most compute wins. Alphabet is betting that $85 billion in fresh equity is cheap compared to the alternative, which is showing up to the AI arms race underfunded.

What is remarkable is that the market took this dilution in stride. Normally when a company announces it is selling tens of billions of dollars in new stock, shareholders run screaming because their ownership percentage is being sliced thin. The fact that investors essentially said yes please and asked for more tells you everything about the moment we are in right now. When the world’s largest equity raise in history gets oversubscribed, the AI gold rush is not a metaphor anymore. It is a balance sheet entry.

DEEPSEEK, THE CHINESE AI THAT RATTLED SILICON VALLEY, IS CLOSING A $7 BILLION FIRST-EVER FUNDING ROUND WITH TENCENT AND STATE MONEY POURING IN

Remember January of last year when DeepSeek dropped a model that rivaled OpenAI’s best work, apparently built for pocket change, and caused a one-day stock market panic that wiped hundreds of billions off American tech valuations? Well, it turns out that Chinese investors remember that day too, and they have decided it is time to write some very large checks.

DeepSeek is closing in on a $7.4 billion funding round, which would be one of the largest startup financings in Chinese history. The company is doing this at a valuation of around $52 billion. Here is the kicker: DeepSeek has never raised outside capital before. This would be their first round ever. They went from zero to $52 billion in valuation without taking a single dollar from outside investors. That is either a sign of extraordinary frugality or the most successful bootstrapped operation in the history of artificial intelligence, depending on how generous you are feeling.

The investor list reads like a who’s who of Chinese big tech and state capital. Tencent is considering a 10 billion yuan commitment. CATL, the battery giant that builds the cells inside most of the electric vehicles on the planet, is looking at 5 billion yuan. The National Artificial Intelligence Industry Investment Fund, which is Beijing’s direct arm for funneling state money into strategic tech, is participating. NetEase and JD.com are in talks to join as well. The founder himself, Liang Wenfeng, is putting in 20 billion yuan of his own money, roughly $2.7 billion. The man is not just building the company, he is buying a huge chunk of it with his personal funds.

This deal matters for a few reasons beyond the headline number. First, it signals that the Chinese government has decided DeepSeek is a national champion worth backing explicitly, not just implicitly through favorable treatment. When the state AI fund shows up in your cap table, you have been officially anointed. Second, the participation of companies like Tencent and CATL means the Chinese corporate establishment sees DeepSeek as infrastructure for the next decade. Third, a $52 billion valuation for a company that cracked the efficiency problem in AI training suggests the market believes this approach is not a one-trick pony.

The strategic context here is that China is trying to build an AI ecosystem that does not depend on American chips or American models. DeepSeek showed last year that world-class AI is achievable with less compute, which matters enormously when Nvidia export controls are limiting what chips China can buy. A well-funded DeepSeek pursuing that efficiency-first approach is a more durable competitive threat than a capital-intensive lab trying to buy its way to the frontier. This is not just a fundraising story. It is the formalization of China’s AI industrial policy through private capital markets, with the state and the biggest tech conglomerates all pointing in the same direction at once.

ANTHROPIC RAISES $65 BILLION, OVERTAKES OPENAI IN PRIVATE VALUATION AT $965 BILLION, THEN IMMEDIATELY FILES FOR AN IPO

There is a running joke in Silicon Valley that the current AI funding environment is so detached from reality that normal financial concepts simply do not apply. Anthropic just provided the single best evidence for that argument by raising $65 billion at a $965 billion valuation and then, six days later, filing to go public. They raised the equivalent of the GDP of a mid-sized country and then decided that was not quite enough and they needed the public markets too.

The round, a Series H, was led by Altimeter Capital, Dragoneer, Greenoaks, and Sequoia Capital. Also involved were Capital Group, Coatue, D1 Capital Partners, Baillie Gifford, Blackstone, Brookfield, D.E. Shaw Ventures, DST Global, and Fidelity. Samsung, SK Hynix, and Micron came in as strategic infrastructure partners, which makes sense because chip companies want to be close to the people training the next generation of models. A chunk of the round, around $15 billion, consisted of previously committed investments from cloud providers including Amazon, which had pledged $5 billion back in April.

The valuation at $965 billion puts Anthropic officially ahead of OpenAI for the first time in the private markets. OpenAI raised at around $300 billion earlier this year. Anthropic just tripled that implied ceiling. The two companies are now in a valuation arms race that has nothing to do with revenue multiples or traditional financial metrics and everything to do with who controls the most capable AI models at scale.

Anthropic’s co-founders are now each worth around $8 billion on paper following the round. Dario Amodei, who left OpenAI in 2021 with a small team and a conviction that safety-focused AI development deserved a serious organization, has built something that is now valued at nearly one trillion dollars. That is a remarkable trajectory by any measure, and the kind of outcome that will be studied in business school case studies for decades.

The IPO filing came on June 1st, making this one of the most compressed fundraise-to-public-markets timelines in recent memory. The sequence of events suggests Anthropic and its bankers decided that the private market would give them one more monster round, and then they would take the rest of what they need from public investors before anyone had time to reassess whether $965 billion is a defensible number for a company competing in a market where every major tech giant is trying to catch up. The window is open right now, and they are walking through it at a sprint.

Read more: TechCrunch | CNBC | Anthropic IPO filing

VIBE CODING TURNS A BORING DATABASE STARTUP INTO A $10.5 BILLION COMPANY AS SUPABASE RAISES $500 MILLION

If you had told someone three years ago that a Postgres database startup would hit a $10.5 billion valuation, they would have handed you a glass of water and suggested you sit down. Databases are not glamorous. They are not the kind of thing that makes venture capitalists write breathless blog posts about changing the world. They are the plumbing. But it turns out that when half of Silicon Valley suddenly decides that AI coding tools are going to let anyone build software by talking to a chatbot, the plumbing companies become very valuable very fast.

Supabase just raised $500 million in a round led by GIC, Singapore’s sovereign wealth fund, with participation from Accel, Y Combinator, Craft, Felicis, Coatue, and Stripe. The company is now valued at $10.5 billion including the fresh capital. That valuation has roughly doubled since October of last year, and it had already doubled before that. They are compounding at a pace that makes most growth-stage companies look like they are standing still.

The story behind the numbers is the vibe coding wave. Co-founder and CEO Paul Copplestone said in the announcement that AI coding tools are now responsible for the majority of new databases being created on Supabase’s platform. Claude Code, Anthropic’s agentic coding assistant, is the single largest driver of new Supabase databases in 2026. When someone fires up Claude Code or a similar tool and asks it to build a web app, the AI reaches for Supabase as a backend by default. It is not magic. It is the result of years of developer-friendly API design and open source community building. But the timing is extraordinary.

Supabase has over 250,000 customers and 350 employees. The company was founded in 2020 as an open source alternative to Firebase, Google’s backend-as-a-service product. The pitch was simple: take a technology that developers already know and love, Postgres, and make it as easy to use as the proprietary alternatives. That bet is now paying off at a scale the founders probably did not anticipate when they were writing the first lines of code.

The broader lesson here is that infrastructure companies that made choices to be developer-friendly and open source before the AI boom are now reaping returns they did not plan for. When AI coding tools pick their defaults, they pick the tools developers already use and trust. Supabase built that trust over years. The $10.5 billion valuation is the market saying that trust has real economic value in a world where AI agents are doing a growing share of the software development work.

NVIDIA IS NOW ONE OF THE BIGGEST AI INVESTORS ON EARTH WITH OVER $40 BILLION IN EQUITY BETS THIS YEAR ALONE

At some point Nvidia stopped being just a chip company and became something harder to categorize. It still makes chips. The chips are still the most important product in the artificial intelligence supply chain. But in 2026, the company also deployed over $40 billion in equity investments across AI startups, public companies, and infrastructure plays. Jensen Huang has turned the most valuable semiconductor company in history into a venture capital operation that also happens to manufacture the hardware everyone in the industry needs to buy.

The numbers are staggering. The single biggest bet was a $30 billion investment in OpenAI, which alone would make Nvidia one of the top investors in private AI by any measure. But the company also put $5 billion into Intel earlier this year, and that position has already grown to over $25 billion as Intel’s stock recovered, making it one of the most lucrative short-term corporate investments in recent memory. Beyond that, Nvidia committed up to $3.2 billion to Corning, the glassmaker that supplies optical fiber for data center interconnects, and up to $2.1 billion to IREN, a data center operator. The company has also participated in around two dozen private startup funding rounds in 2026 alone.

The strategic logic is not subtle. Nvidia’s chips power the AI ecosystem. If the AI ecosystem thrives, Nvidia sells more chips. By taking equity stakes in the companies that buy the most chips, Nvidia is hedging in multiple directions simultaneously. If one of these companies becomes wildly successful, Nvidia captures some of that upside directly on top of the hardware revenue. If the broader ecosystem grows, every company in the portfolio buys more compute. It is the picks-and-shovels gold rush strategy taken to its logical extreme, executed with a balance sheet that most sovereign wealth funds would envy.

There is also a signaling dimension here. When Nvidia puts $30 billion into OpenAI, it is not just a financial bet. It is a public declaration that Nvidia considers OpenAI the most important customer and partner in its ecosystem. That kind of institutional endorsement changes how other companies and investors think about OpenAI’s prospects. The investment is simultaneously financial, strategic, and reputational.

The question going forward is whether regulators start paying attention to a chipmaker that also owns significant equity stakes in most of the major AI labs it sells hardware to. That kind of vertical integration in an industry this strategically important has a tendency to eventually attract uncomfortable scrutiny. For now, though, Nvidia is playing every angle of the AI economy at once and winning on all of them, which is either a masterclass in corporate strategy or a conflict of interest waiting for someone in Washington to notice.