🔥📊 Yesterday, both Buffett and Duan Yongping adjusted their portfolios in sync: The AI theme is becoming "institutionalized and concentrated."

In the latest disclosed holdings from yesterday, there's a signal more important than price movements—

Big money is converging, not dispersing.

I'll talk about the commonalities first, then the divergences.

First, commonality:

Both Buffett and Duan Yongping hold Google.

This in itself isn't surprising, but in the current AI competitive landscape, it looks more like a bet on "underlying computing power + data gateway."

Google isn't just search; it's:

AI foundational models

Cloud computing

An advertising cash flow machine

It belongs to "deterministic infrastructure assets," not high-volatility themes.

Second, commonality:

Both reduced their positions in Apple, but it remains their largest holding.

This looks more like a structural rebalancing, not a trend of bearishness.

Apple is already a super cash flow asset; reducing the position is more about portfolio management than a shift in logic.

What's really worth dissecting is the divergence.

Duan Yongping significantly increased his position in $NVIDIA(NVDA.US), directly elevating it to his third-largest holding, just behind Apple and Berkshire.

This isn't a tactical trade; it's a strategic bet.

When a long-term investor who emphasizes "circle of competence" actively pulls Nvidia into his core portfolio tier, it shows he already sees it as a representative of "AI infrastructure pricing power."

Looking at the new additions:

$Coreweave(CRWV.US)

$ASML(ASML.US)(ASML.US)(ASML.US)

$Tempus AI(TEM.US)

The commonality of these three is: they are all deeply related to the AI or semiconductor chain, but at different levels—

$ASML(ASML.US)(ASML.US)(ASML.US) is the equipment leader

$Coreweave(CRWV.US) is related to computing power/cloud

$Tempus AI(TEM.US) leans towards medical AI data

This isn't random position building; it's a layout around the upstream and application ends of AI.

Looking at the institutional level.

Bridgewater and Baillie Gifford also increased their positions in $NVIDIA(NVDA.US).

When two types of institutions with vastly different styles simultaneously increase their bets on the same asset, it usually indicates:

Not emotion-driven, but trend confirmation.

They aren't betting on "how much it will rise," but on "the sustainability of AI capital expenditure."

SoftBank's moves are even more contrasting.

Q4 liquidation of $NVIDIA(NVDA.US) was interpreted by the market as a shift of funds towards OpenAI.

But at the same time, they built a position in Circle.

This represents another line of thinking—

Not betting on computing power, but betting on "AI + financial infrastructure."

Circle is a stablecoin issuer, essentially the hub for on-chain US dollar liquidity.

If the AI economy expands, payment and settlement infrastructure will also become foundational support.

So what you're seeing now isn't "who is bullish on AI," but:

Funds are beginning to stratify within the AI track.

First layer:

Deterministic computing power leader — $NVIDIA(NVDA.US)

Second layer:

Foundational platform — Google

Third layer:

Equipment chain — $ASML(ASML.US)(ASML.US)(ASML.US)

Fourth layer:

Application and derivatives — $Tempus AI(TEM.US), $Coreweave(CRWV.US)

Fifth layer:

Financial infrastructure — Circle

The real question isn't "whether to participate in AI," but:

Do you value computing power pricing power more, or the speed of application diffusion?

More crucially—

When institutions are already highly concentrated in Nvidia and Google, is this consensus a safety cushion, or a future crowded trade?