That’s a $21B "conviction only" quarter from Jensen 👀#COWRises55.77%In24h
*NVIDIA Q2 2026 13F breakdown:*
- *Portfolio*: $63.44B → *+245% QoQ*
- *But*: Strip out mark-to-market + SpaceX and *zero new buying*
- *The move*: *$21B into SpaceX* = ∼122.8M Class A shares. Now *33.1% of the book*
- *Top 2 holdings*:
1. *$INTC*: 47.3%
2. *SpaceX*: 33.1%
- *Everything else*: $CRWV, $COHR, $NOK, $SNPS, $NBIS = *exact same share count as Q1*. No trims, no adds
*What this tells you:*
1. *"Hold and let it run" playbook*
Jensen isn’t rotating. He’s letting AI winners compound. If you were right in 2024-2025, just hold
2. *One massive swing: SpaceX*
This isn’t about chips. This is about *AI + infrastructure + compute in orbit*. Starlink data, satellite AI, low-latency inference. NVIDIA is betting the next frontier of AI infra is space
3. *Core conviction names intact*
Intel still #1 holding. Even with all the "Intel is dead" talk, NVDA didn’t sell a share. Same for the AI infra picks: $CRWV, $SNPS, $NBIS
4. *No exits*
Usually when a portfolio 3x’s you see profit-taking. NVDA did none. That’s peak confidence
*Why SpaceX matters for NVDA thesis:*
NVIDIA sells the picks and shovels. But if compute moves off-planet, they need to own the platform too. SpaceX = global data + compute + distribution.
This lines up with their "AI factories" + "robotics + sovereign AI" narrative.
*Watch next quarter:*
- *Do they trim $INTC*? 47% in one legacy name is risky long term
- *Any new "infrastructure" adds*? Energy, robotics, or more space bets
- *$CRWV/$NBIS*: If those run, NVDA’s book gets even more concentrated in AI infra
*Bottom line:*
This wasn’t a trading quarter. It was a statement quarter.
$21B into SpaceX says: "AI training is solved, now we need AI everywhere — including space."
Clean, patient, and now extremely concentrated. If SpaceX IPOs or does another big round, that $63B book could 2x again on paper.
#BNBChainToActivatePasteurHardFork
*NVIDIA Q2 2026 13F breakdown:*
- *Portfolio*: $63.44B → *+245% QoQ*
- *But*: Strip out mark-to-market + SpaceX and *zero new buying*
- *The move*: *$21B into SpaceX* = ∼122.8M Class A shares. Now *33.1% of the book*
- *Top 2 holdings*:
1. *$INTC*: 47.3%
2. *SpaceX*: 33.1%
- *Everything else*: $CRWV, $COHR, $NOK, $SNPS, $NBIS = *exact same share count as Q1*. No trims, no adds
*What this tells you:*
1. *"Hold and let it run" playbook*
Jensen isn’t rotating. He’s letting AI winners compound. If you were right in 2024-2025, just hold
2. *One massive swing: SpaceX*
This isn’t about chips. This is about *AI + infrastructure + compute in orbit*. Starlink data, satellite AI, low-latency inference. NVIDIA is betting the next frontier of AI infra is space
3. *Core conviction names intact*
Intel still #1 holding. Even with all the "Intel is dead" talk, NVDA didn’t sell a share. Same for the AI infra picks: $CRWV, $SNPS, $NBIS
4. *No exits*
Usually when a portfolio 3x’s you see profit-taking. NVDA did none. That’s peak confidence
*Why SpaceX matters for NVDA thesis:*
NVIDIA sells the picks and shovels. But if compute moves off-planet, they need to own the platform too. SpaceX = global data + compute + distribution.
This lines up with their "AI factories" + "robotics + sovereign AI" narrative.
*Watch next quarter:*
- *Do they trim $INTC*? 47% in one legacy name is risky long term
- *Any new "infrastructure" adds*? Energy, robotics, or more space bets
- *$CRWV/$NBIS*: If those run, NVDA’s book gets even more concentrated in AI infra
*Bottom line:*
This wasn’t a trading quarter. It was a statement quarter.
$21B into SpaceX says: "AI training is solved, now we need AI everywhere — including space."
Clean, patient, and now extremely concentrated. If SpaceX IPOs or does another big round, that $63B book could 2x again on paper.
#BNBChainToActivatePasteurHardFork