The Dow's Biggest Record Day Of The Year Had Nothing To Do With AI. That's The Best News In This Paper.

For three years the bull market has been accused of standing on one leg: artificial intelligence. This week the market showed everyone what the second and third legs look like. Health care led — UnitedHealth Group (UNH) rose 5.2% on a Bank of America upgrade and raised its dividend. Financials had their best single day since April 2025: Goldman Sachs (GS) and Citigroup (C) hit 52-week highs, JPMorgan Chase (JPM) added more than $10 a share, and the small-cap Russell 2000 outpaced everything at +1.45%. Meanwhile the semiconductor index dropped 2.15%. A year ago, a chip selloff of that size would have dragged the whole market down with it. On Thursday, the market shrugged and set a record anyway.

This is the rotation we have been positioning for all spring — not by abandoning the AI infrastructure trade (we still believe in it, and Nvidia (NVDA) rose 1.9% on the day, decoupling from Broadcom's stumble), but by refusing to let it be the only engine in the book. The Dollar General (DG) add last Wednesday, the banks, the energy overweight, the gold sleeve — that's the other leg. Weeks like this are why it exists.

We made two additions this week — Macy's (M) and UnitedHealth Group (UNH) — both straight out of the rotation playbook, and we put Broadcom (AVGO) on a tightened leash. The full reasoning is below, and the cascaded positions are on the Portfolios page.

Broadcom Lost $286 Billion In A Day — For Meeting Expectations. Here's What We're Doing About It.

Here is what Broadcom (AVGO) did wrong on Thursday: nothing, exactly. The company affirmed its forecast of more than $100 billion in AI-chip sales for next year. Its quarterly AI revenue of $10.8 billion came in just barely above its own guidance, and next quarter's $16 billion outlook landed slightly below Wall Street's hopes. For that, the stock fell 13% and erased $286 billion of market value — the fourth-largest one-day decline ever recorded for a U.S. company. The leveraged Broadcom ETFs — yes, those exist — fell 25%.

When "as expected" costs a company a quarter-trillion dollars, the message isn't about the company — it's about the price. The most expensive corners of the AI trade are priced for acceleration, and merely excellent is no longer good enough. There were real wrinkles in the report worth respecting: Anthropic reportedly shifted some orders from Broadcom's full server racks to chips only, and there's persistent chatter about Google diversifying its chip suppliers. But Morgan Stanley still estimates Broadcom keeps at least 80% of Google's custom-chip business, and the memory-and-contagion selloff that dragged down Micron (MU) and AMD (AMD) spared Nvidia (NVDA) entirely — up 1.9% on the day.

We are not selling Broadcom on one guidance wobble — the order book is real and enormous. But we're done adding at these prices, and we trimmed the position back to target where the rally had let it swell. Discipline means letting winners run and cutting them back to size when one day's mood swing is worth $286 billion. Full Broadcom note here.

Private Credit's "Anything Goes" Era Just Ended. Investors Are Heading For The Exits — Slowly, Because They Have To.

Friday's Journal put it on the front page: the private-credit industry — the $3 trillion business of non-bank lending that boomed when money was easy — is tightening fast. Since March, lenders have been raising rates, fattening fees, capping leverage and closing the loopholes that let troubled borrowers strip collateral away from their creditors. The median margin on new large direct loans has climbed three months straight, to its highest level in nearly two years. As Ares (ARES) chief executive Michael Arougheti put it, the market has flipped "from a borrower-friendly market to a lender-friendly market." Translation: the easy-money era is over, and the lenders know something.

The other half of the story is the exit line. Investors asked to redeem 10% of Blackstone's (BX) $79 billion Bcred fund in the second quarter — about $4.4 billion — and Blackstone capped the payout at 5%, reversing a promise from March to pay redemptions in full. At Cliffwater's $31 billion fund, withdrawal requests hit 17%. Blue Owl (OWL) faced 22% earlier this year. These funds aren't failing — they're working as designed. But "working as designed" means your money comes out five percent per quarter when the crowd heads for the door, and plenty of investors are just now reading that part of the brochure.

Our models hold zero private-credit funds and zero BDCs — we never reached for the extra yield that's now gated. Our cash buffer is short-duration Treasuries, which cannot be gated by anyone. This story is the entire reason why. We updated our full Credit Stress 2026 scenario briefing with this week's numbers — it's now interactive. The full private-credit read here.

Buffett's Team Went Shopping At Macy's. We Followed Them In.

Quietly, in a regulatory filing, Berkshire Hathaway (BRK.B) disclosed a stake in Macy's (M) — yes, the parade-and-perfume-counter Macy's — and the stock is up roughly 25% since. Here's why the smartest value shop on earth is interested in a department store: Macy's trades at about 1.3 times book value and 11 times forward earnings, pays a 3.2% dividend, and — this is the part that matters — analysts peg its real estate at $7.9 to $10.5 billion, comfortably more than the entire company's market value. You're buying the buildings and getting the business for free.

And the business itself has stopped shrinking — Thursday's paper carried the proof. Macy's reported quarterly sales up 1.8%, comparable sales up 3% (the fourth straight gain), profit up 66%, and raised its full-year guidance, with Bloomingdale's comps up 10.2% and competitor Saks closing nearly a third of its Saks Fifth Avenue and Neiman Marcus stores — handing market share to the survivor. On the day the Dow set its record, this is the kind of unloved, asset-backed, dividend-paying name that led. It's also precisely the style of investing — price paid determines return — that we preach from the Graham-and-Buffett hymnal every week.

We added Macy's (M) as a starter ~1% value position in the income-tilted tiers, alongside our existing Berkshire (BRK.B) anchor. The full Macy's case — including the real-estate math — here.

Health Care Led The Dow To Its Record. We Added The Stock That Did The Leading.

UnitedHealth Group (UNH) rose 5.2% on Thursday after a Bank of America upgrade, led the entire Dow to its record close, and raised its quarterly dividend to $2.32 a share. After a brutal stretch for managed care, the biggest health insurer in America is being re-rated — and health care as a sector is doing what it historically does best: taking the baton when expensive growth stumbles. On a day when semiconductors fell more than 2%, health care and financials carried a +875-point Dow record. That is textbook late-cycle broadening.

We like UnitedHealth here for three plain reasons. First, valuation: it spent a year in the penalty box and still trades far below its historical premium. Second, the dividend raise is management's way of saying the worst is behind it. Third, portfolio math: our book has been deliberately adding non-AI engines — consumer staples via Dollar General (DG), banks, energy, and now health care — so the models don't live and die on one theme's Tuesday mood.

We added UnitedHealth Group (UNH) at ~1.5% in the core and dividend tiers. One honest caveat: managed care is a regulatory football in an election year — which is exactly why the position is sized as a starter, and why we updated the Midterms 2026 briefing this week (it's now interactive, and worth five minutes).

Oil Dipped On A Ceasefire. Hours Later, A Drone Hit Kuwait's Airport. The Inventory Clock Is Still Ticking.

West Texas crude (WTI) slipped 3.1% to $93.04 on Thursday as a renewed Israel-Lebanon ceasefire raised hopes the region might cool. The market's optimism lasted about as long as the headline: an Iranian drone struck the main terminal at Kuwait's international airport — the third attack on Kuwait in about a week — and the Strait of Hormuz remains closed. Gasoline is up 35% from a year ago, diesel 53%. And yet Americans drove a record 280 billion miles in April. Demand is not blinking — though it is adapting: Thursday's paper reported hybrid-vehicle sales jumped 33% in May as drivers flee $4.26 gasoline, the quiet winner being Toyota (TM), which we hold.

The piece of Friday's paper we'd make every client read is the Heard on the Street column calling this "the largest energy-supply shock ever" — and quoting Exxon and Chevron executives warning that global inventories reach a critical point within weeks, just as summer driving season peaks. Even if a deal reopened Hormuz tomorrow, clearing mines and re-insuring tankers would keep flows slow for months. One veteran analyst put it bluntly: "red lights are going to start flashing this month."

This is why the energy overweight — ExxonMobil (XOM), Chevron (CVX), Cheniere Energy (LNG) — doesn't move on a ceasefire headline. We reinforced all three. The full oil-inventory note here.

The Big Banks Are Building Their Own Crypto Rails — And The Hated $25,000 Day-Trading Rule Just Died.

Two structural stories for the financials sleeve landed on the same day its stocks had their best session in over a year. First: the largest U.S. banks — JPMorgan Chase (JPM), Bank of America (BAC), Citigroup (C) and Wells Fargo (WFC) — are planning a shared tokenized-deposit network through The Clearing House, targeting launch in the first half of 2027. In plain English, they're putting ordinary bank deposits on blockchain rails that settle around the clock — a direct counterattack on the stablecoin companies trying to siphon deposits out of the banking system. The banks aren't fighting crypto anymore; they're absorbing it.

Second: Thursday was the last day of the 25-year-old "pattern day trader" rule, which required $25,000 of account equity to trade actively. Robinhood (HOOD) rose 5.5% and lifted the retail-brokerage complex with it — both Robinhood and Webull jumped more than 10% back in April when the change was approved. More trading volume is a structural tailwind for the brokers, whatever you think of the wisdom of more day trading (we'll keep coaching the opposite).

We reinforced JPMorgan Chase (JPM), Goldman Sachs (GS) and Morgan Stanley (MS) — the moat just got deeper and the fee pipeline (SpaceX, OpenAI and Anthropic IPOs are all queued) keeps the engine hot. Robinhood (HOOD) goes on the watch list, not in the models. Full tokenized-deposits brief here.

Washington Wants A Piece Of The AI Boom. The AI Boom Wants A Breather. Both Sentences Are Real.

Friday's paper carried two AI stories that would have been science fiction eighteen months ago. One: senior U.S. officials have discussed the federal government taking direct equity stakes in major AI companies — an idea pitched by OpenAI's Sam Altman himself — while Senator Bernie Sanders drafts legislation that would transfer 50% of top AI companies' equity into a public fund. The administration already holds direct investments in more than ten companies, including Intel (INTC). Two: Anthropic — which just closed a funding round near a $1 trillion valuation and is on track for $50 billion in annualized revenue, up from $9 billion at the end of 2025 — publicly called for the leading labs to consider slowing down, warning its own models are on a path toward "recursive self-improvement."

Put the two together and you get the theme we've been tracking since the Trump AI executive order: the AI industry is becoming a regulated, state-entangled utility in real time — more like aerospace and banking, less like a Silicon Valley free-for-all. For investors, that's not bearish; it's clarifying. Regulation favors incumbents with compliance budgets — Microsoft (MSFT), Alphabet (GOOGL), Nvidia (NVDA) — and government equity, if it ever happens, would make Washington a stakeholder in the boom continuing. Meanwhile the third data point from Friday's paper deserves its own sentence: AI is now the No. 1 cited reason for U.S. layoffs, three months running, at 22% of all job cuts this year.

No position changes on the politics alone — but it reinforces the architecture: infrastructure over applications, balance-sheet giants over borrowed-money startups, and a gold sleeve (IAU) for the institutional-upheaval scenarios nobody can price. The full Washington-and-AI read here.

America Has The Money For Its Data-Center Boom. It Can't Find The Transformers.

Thursday's Journal led with the most important AI story nobody trades on: the capital keeps growing, but the construction can't keep up. A JPMorgan analysis found that more than 60% of the data-center capacity planned for 2027 completion isn't under construction yet, with another 7% already delayed — held back by gas-turbine and transformer backlogs, permitting fights, and simple power availability. Microsoft (MSFT), Alphabet (GOOGL), Meta (META) and Amazon (AMZN) spent $410 billion on capital projects last year and are expected to top $670 billion this year. The money is not the constraint. Electricity is.

The same day, Alphabet announced an $85 billion equity raise to fund the build-out — with Berkshire Hathaway (BRK.B) taking $10 billion of it at a negotiated discount, Greg Abel's second deal in a single weekend. And here's the detail that makes Alphabet the class of the field: it's the only big tech company that owns a power company, having bought renewables developer Intersect for $4.75 billion, and it just signed a deal to free up 100 megawatts in the country's largest power market. When the bottleneck is electricity, the company with its own utility wins.

This is the thesis we've been pounding for months — own the picks, shovels and power of the AI build-out, not the apps. We reinforced Alphabet (GOOGL), already in the book, with the Berkshire co-sign as confirmation. The full bottleneck brief here.

Bitcoin's Winter Turns Bitter While SpaceX Prices A $1.75 Trillion Debut. Watch Where The Air Goes.

Two stories from Thursday's paper, same lesson. First: bitcoin fell a fourth straight day to about $65,380 — down 25% this year and roughly half its October peak — after Michael Saylor's Strategy (MSTR) sold bitcoin for the first time since 2022 to pay a preferred dividend (the stock fell 7%). The twelve spot-bitcoin funds have bled $4 billion over a 12-day outflow streak — the first true bear market of the ETF era. We've said it since the Saylor sale broke: a company whose only engine is a volatile asset must sell that asset at the worst moments. We own none of it.

Second: SpaceX set terms for its market debut — 555,555,555 shares (Elon being Elon) at $135 each, raising roughly $75 billion at a ~$1.75 trillion valuation, with trading expected June 12. That's about 94 times revenue for a company that lost $4.9 billion last year — and it would be the largest listing of all time. Here's why both stories matter to a book that owns neither: the biggest IPO ever and a crypto bear market are competing claims on the same speculative dollar. One crypto trader put it perfectly: bitcoin's recovery needs "some of the air coming out of the AI trade." Watch the flows, not the headlines.

No positions in bitcoin, MSTR or the SpaceX debut — but Goldman Sachs (GS) and Morgan Stanley (MS), which we reinforced this week, are fee-collectors on the entire IPO calendar regardless of where it prices. The house always gets paid. The full crypto-winter note here.

Ta-Da! Scientists Take Improv Class, A Man Hunts 50-Year-Old Candy, And The Knicks Refuse To Lose.

The best thing on Friday's front page wasn't a market story. It was a neuroscientist in an improv class pitching an imaginary "hammer humidifier" — "the top-rated humidifier in Arizona… and it can be used for self-defense!" — because scientists across the country are taking comedy classes to learn how to talk to the rest of us. Alan Alda's science-communication center has trained 35,000 of them, funded in part by auctioning his M*A*S*H boots for $125,000. Our favorite detail: a chronic-pain researcher who now quietly whispers "Ta-da!" to herself whenever something goes wrong. Honestly, that's not bad investment advice either. The full story →

Elsewhere in the paper: an op-ed writer spent decades hunting Choc-O-Mint Life Savers (discontinued 1981), found a 50-year-old pack sealed in a Mason jar from a Colorado barn sale, and ate one over his wife's objection ("Are you insane?"). Verdict: "deliciously peppy." The Mansion section coined "med-à-terre" for retirees buying apartments near their cardiologists instead of condos near the beach — there's a genuinely useful retirement-planning lesson in that one. And the Knicks beat the Spurs in Game 1 of the Finals behind a hobbled Jalen Brunson's 30 points; they haven't lost a game in six weeks and are chasing their first title in 53 years.

Have a great weekend. The weekend edition lands Saturday; bring coffee.