The single most interesting signal this quarter: top funds split on Alphabet

Two of the most-watched fundamental investors looked at the same stock and went opposite directions.

✅ Berkshire Hathaway (Greg Abel's first major print) loaded up on GOOG:

• Position: USD 5.59 billion → USD 16.63 billion
• Class A shares: +204%
• Weight: 2.0% → 6.3% (now a core holding)

❌ Pershing Square (Bill Ackman) almost exited GOOG:

• Position: USD 2.147 billion → USD 99 million
• Share count: -95%
• The freed capital went to a brand-new Microsoft position (USD 2.09 billion, 15.3% weight)

When two legends disagree this sharply on the same name, it tells me how concentrated and contested AI mega-cap conviction has become.

The 5-fund snapshot

1. Berkshire Hathaway: the Abel era opens with a GOOG bet

• Added: GOOG (large), NYT, Lennar
• New: Delta Air Lines (USD 2.65 billion)
• Trimmed: Chevron, Constellation
• Exited: Amazon, Visa, Mastercard
• Signal: GOOG just became a core position. That's a tell.

2. Pershing Square: same tech weight, different ticker

• New core: Microsoft (USD 2.09 billion, 15.3%)
• Cut: Alphabet by 95% of shares
• Increased: Amazon +19.2%
• Trimmed: Brookfield, Uber
• Exited: Hilton
• Signal: Tech allocation didn't shrink. It rotated.

3. Bridgewater (Ray Dalio): shrank the book, doubled down on AI compute

• Total disclosed AUM: USD 27.42 billion → USD 22.40 billion (-18.3%)
• AI / chip clearly up:
  • Amazon +125.3%
  • NVDA +21.4%
  • AVGO +57.4%
  • Micron added
  • TSM new position (USD 364 million)
• Trimmed: IVV -36%, SPY -2%
• Exited: Salesforce, Workday, ServiceNow
• Signal: Less blended index exposure, more direct AI compute supply chain.

4. Duquesne (Stanley Druckenmiller): concentrated, smaller, biotech + EM energy

• Total disclosed AUM: USD 4.494 billion → USD 3.377 billion (-24.9%)
• Top holding: Natera (USD 575 million → USD 613 million, weight 12.8% → 18.1%)
• YPF (Argentina oil & gas): USD 22 million → USD 150 million (+433% in shares)
• Added: RSP calls, Broadcom, ARGT (Argentina ETF)
• Increased: Natera, YPF, BBB Foods
• Exited: XLF (financials), Alphabet, RSP common shares
• Signal: Trim the book, concentrate on biotech genetics and emerging-market energy.

5. Soros Fund Management: still long AI, heavy index hedges

• Total disclosed AUM: USD 8.631 billion → USD 9.119 billion (+5.7%)
• AI / tech longs up: NVDA +61.2%, TSM +49.3%, Electronic Arts +125.4%
• SPY puts: USD 102 million → USD 773 million (+658% on the put dollar value)
• SPY put weight: 1.2% → 8.5%
• Added: Linde, TSM puts
• Trimmed: Amazon, Microsoft, Salesforce
• Exited: Spotify convertibles, Confluent, Dropbox
• Signal: Portfolio got bigger, but the defensive hedge got materially bigger faster.

The under-the-radar fund worth knowing: Situational Awareness LP

Run by Leopold Aschenbrenner, a former OpenAI researcher who flagged AGI as the dominant capital allocation theme of the decade.

The fund's Q4 2025 13F (positions as of 31 Dec 2025) was its first public filing. Q1 2026 hasn't been filed yet.

If you had hypothetically held those Q4 2025 disclosed common stock positions from 31 Dec 2025 to 14 May 2026 (~4.5 months):

• Common stock only return: +147.8%
• QQQ same window: +17.2% (614.31 to 719.79)
• Outperformance vs QQQ: +130.7 percentage points

Including 13F-disclosed call/put positions on a nominal-approximation basis, return lands closer to +143.9% (outperformance ~+126.7 pp). The clean apples-to-apples figure is the common-stock-only number.

Top contributors: BE, INTC (call options), SNDK, LITE, CRWV, CORZ, APLD.

Caveat that matters: 13F doesn't disclose option strikes, expiries, or actual prices. The option-included return is a directional approximation only. The cleanest read is the common-stock figure.

🔎 Two 13F limitations I always read first

1. Lag. Today's filings show positions as of 31 March. That's ~45 days old. April and May moves may already have reshaped the book.
2. Coverage. 13F shows US long equity only. No shorts. Limited derivatives disclosure (only long puts and calls). No non-US assets. A book that looks aggressively long on paper may be hedged elsewhere.

A "bullish" 13F is not a buy signal. It's a positioning snapshot from 6+ weeks ago.

👀 What I'm watching:

1. Is AI capex still in the institutional core? Berkshire's GOOG add and Bridgewater's NVDA / AVGO / AMZN / MU stack say yes.
2. Is AI exposure broadening past pure-play GPUs? Bridgewater into TSM (foundry), Soros into TSM, Duquesne into AVGO. The trade is widening to include foundry + networking + downstream infrastructure.
3. Where's the defence? Soros' SPY put line is the cleanest tell. Think of it as portfolio insurance: the put dollar value rises only if SPY drops.
4. For SG investors in CSPX, IWDA, or VWRA: You already own most of the Berkshire and Pershing top names indirectly. The 13F is more useful for sizing decisions than for adding new tickers.