The Berkshire endorsement is a rare contrarian data point
for the AI-bubble framework. The world's most valuation-
disciplined equity investor making Alphabet its third-largest
position — after having sat on record cash and net-selling for
14 straight quarters — weakens the pure "peak-AI-mania" thesis.
It doesn't disprove the bubble reading (see our
historical-lens digest),
but it does reprice one variable: Alphabet's ~30–35× fwd P/E is
apparently inside Berkshire's margin of safety.
The Alphabet-as-utility framing is now official.
Berkshire's late-2010s tech shift centered on Apple as "consumer-
staples-in-a-pocket." The 2026 Alphabet accumulation looks like
the same playbook applied to search + cloud + YouTube +
Waymo: a moat-quality, cash-generative utility with an
AI-compute super-cycle lifting all three legs. Pichai's $180–190B
2026 capex is the CAPEX phase; Berkshire is underwriting the
subsequent cash flows.
The 14-quarter streak breaking is the bigger news.
A single position add — even $17B — is one data point. Ending a
multi-year net-selling regime is a regime signal. Every Berkshire
read for the next 12 months should reference this pivot. Under Abel,
Berkshire is done being defensive.
Concentration risk cuts both ways. Top-4 = ~62% of
the ~$310B disclosed equity book. That's a lot of eggs in AAPL, AXP,
GOOG, KO. If AI-compute economics don't play out as priced, Alphabet's
re-rate risk is now Berkshire's re-rate risk — and by extension every
index and every retail portfolio piggy-backing Berkshire's implied
blessing.
Michael Burry filed puts against NVDA + PLTR in the same
window — a much smaller signal in dollar terms, but the two
disclosures landing simultaneously frame the peak-AI debate: Buffett
(via Abel) leaning long Alphabet at ~30× fwd P/E; Burry short the
highest-multiple leaders. Both can be right at once — one holding
the moat-utility, the other shorting the priced-for-perfection tail.
