The $34 Billion 13F Nobody Reads: Gates Foundation Trust Exits Microsoft and Buys a One-Month-Old Spinoff

Hedge fund 13Fs get the coverage. Endowment and foundation 13Fs get ignored — which is odd, because they are usually the cleaner signal. A foundation trust has no performance fee, no redemption risk, no quarterly investor letter to defend. It has one hard constraint: it must fund grants, forever, which means it is a permanent, mandatory seller of something. What it chooses to sell, and what it buys with the proceeds, is about as close to unconstrained preference as public filings get.

The Bill & Melinda Gates Foundation Trust's Q2 2026 filing is a good example, and almost nobody read past the headline.

What the filing actually showed

Three things, in order of how much they matter.

The Microsoft exit is now complete. The trust sold its remaining 7.7 million Microsoft shares in Q1 2026, and the Q2 filing confirms the position at zero. This was a phased liquidation years in the making, not a call on Microsoft's business — the position originated as founder stock gifted to the trust, and a portfolio holding one megacap because of who founded it is a governance problem, not an investment thesis. Still: the single most identity-defining holding in the portfolio is gone.

Berkshire got cut 14%. The trust sold 2.4 million B shares, leaving roughly 14.7 million. Market value fell from about $8.17 billion to $7.35 billion — a reduction of roughly $819 million. Berkshire is still the largest position at around 21% of a $34.42 billion portfolio, so this is trimming, not exiting. But it is the second consecutive signal that the trust is loosening its two historic anchors at the same time.

The proceeds went somewhere unglamorous. A brand-new $352.7 million Home Depot stake (1.0 million shares), and a $180 million position in FedEx Freight Holding (1,192,181 shares). Total portfolio: 24 holdings, up from 22, worth $34.42 billion versus $31.67 billion a quarter earlier. The top five are Berkshire, Caterpillar, Canadian National Railway, Waste Management and Deere.

Read that list again. In a quarter when the entire institutional conversation was about AI capex, the largest philanthropic trust in the country was buying home improvement retail and less-than-truckload freight, and its top five holdings are a conglomerate, two industrials, a railroad and a garbage company.

The FedEx Freight detail is the interesting one

FedEx Freight only began trading independently on June 1, 2026. The trust's position appears in a filing dated June 30. That is a position opened within roughly four weeks of the spinoff's first trade.

This is a category of 13F entry that most screens handle badly. Spinoffs show up as new positions in the quarter they list, which makes them look identical to a fresh conviction buy. Sometimes they are. Often they are mechanical — an index fund or a passive sleeve receiving shares automatically because it owned the parent. Distinguishing the two is the whole game, and the tell is usually share count versus the distribution ratio: if a filer's new stake is roughly what the spinoff formula would have handed them from their old parent position, it's mechanical. If it's materially larger, or the filer never held the parent, someone made a decision.

Spinoff windows are also structurally interesting for a different reason. Newly separated companies get sold indiscriminately in their first weeks by holders who never wanted them — index funds forced out by mandate, parent-company shareholders who own it by accident. That creates a reliable pocket of price-insensitive selling, and patient capital has harvested it for decades. A permanent-capital trust stepping in within a month of listing fits that pattern precisely. You can screen new positions separately from adds on the InvestorLens flow view, which is where these show up first.

Why the boring portfolio might be the point

There's a temptation to read Home Depot, Caterpillar, CN Rail, Waste Management and Deere as timidity. A better reading: it's a portfolio built to fund obligations across decades regardless of which technology narrative wins. Freight, rail, waste and equipment are levered to physical throughput — including, incidentally, the physical buildout underneath the AI trade. Data centers need dirt moved and things shipped.

That's worth cross-checking against how other filers are positioned on growth versus real-economy exposure, which is what the macro consensus view is for. And if you want to see whether the Home Depot and freight entries were idiosyncratic or part of a cluster, the overlap tool will tell you faster than reading twenty filings — one buyer is noise, six independent buyers in a quarter is a rotation. The full filer list lives in the investors directory.

The usual caveats, which bite harder here

Q2 13Fs reflect June 30 positions and were filed in mid-August. It is now September. The trust may have added to Home Depot or reversed the Berkshire trim entirely; Q3 filings aren't due until mid-November.

And the specific caveat for foundation trusts: their selling is partly non-discretionary. A trim can mean "we like this less" or it can mean "we owed $800 million in grants and this was the most liquid thing we owned." A 13F cannot tell you which. Treat the buys as signal and the sells as ambiguous — that asymmetry applies to every endowment filing you'll ever read.

---

Data sourced from public SEC 13F filings. Educational research only — not investment advice.