Thread by @BlackPantherCap

Leopold Aschenbrenner, 24-year-old ex-openai researcher turned $225M into $5.52B in 12 months.

I’ve analyzed all his steps, letters, and filings to get his ‘Playbook.’

🧵Here’s what I think he’s adding:

1/6. The playbook in one sentence:

Find the physical constraint that capital cannot route around. buy the company that controls it. ignore everything else.

Here’s what most people miss about his strategy.

He never bought the names everyone told you to buy.

Instead he asked one question:

“What PHYSICAL constraint can capital not route around?”

Then he bought the companies that control that constraint.

That’s the entire playbook.

What that looked like in practice:

$SNDK entered around $36. Surged to $1,406. Fund’s stake hit $1.48B.

$LITE Lumentum new buy Q4 2025. nvidia committed $2B in procurement. up 122% since.

$CORZ Core Scientific built over 4 straight quarters. share count grew 6.4x. that’s not trading. that’s CONVICTION.

$BE Bloom Energy $875M position. solid-oxide fuel cells that power data centers off the grid entirely, bypassing 3-5 year utility queues.

2/6. Why Bloom over Vistra?

Look at his 13F history quarter by quarter.

Q4 2024: $254M. 6 names. All power infrastructure. Vistra, Constellation, Vertiv. zero chip names.

Q1 2025: Added massive $INTC calls when everyone else was panic-selling on deepseek.

Q3 2025: EXITED Nvidia and Broadcom. both at highs. Called them “consensus.”

Q4 2025: $BE Bloom Energy jumps to #1 at $875M. Lumentum added. ‘Miners’ expanded.

Each rotation follows the same logic:

Find where the bottleneck moved.

He owned Vistra in 2024. Exited it completely in Q4 2025.

The difference: Vistra is grid-connected. bloom is off-grid.

When utility interconnection queues stretch 3-5 years, off-grid modular power isn’t a nice-to-have. It’s the only path to getting a gigawatt cluster online on schedule.

He saw the bottleneck move one level upstream. sold the old position. sized into the new one.

3/6. His Q1 2026 13F drops this week.

Based on running his exact framework forward against what changed in Q1.

Here’s my prediction on what he MOVED.

New position positions:

$COHR — Coherent corp
In march 2026, nvidia committed $2B to lumentum AND $2B to coherent.
Jensen Huang called them partners for “gigawatt-scale AI factories.”
Aschenbrenner already owns Lumentum.

He understands the optical networking bottleneck better than anyone.

Coherent is the direct complement and nvidia just locked the supply chain.

If he didn’t add $COHR in Q1, i’d be surprised.

$GEV — GE Vernova
Transformer lead times are 128-144 weeks right now. Two thirds of planned AI data center capacity is delayed, not because of chips, but because of ELECTRICAL EQUIPMENT.

GE Vernova controls grid-to-chip infrastructure. $52B revenue target by 2028. record backlog. This is exactly the kind of overlooked chokepoint he hunts.

$WULF — Terawulf, low-carbon, nuclear and hydro-powered mining sites. co-led a $1B round for fluidstack, which has a $6.7B 10-year hosting agreement with terawulf starting 2026. This is exactly the kind of pre-priced structural position he builds.

$OKLO — Advanced fast reactors for hyperscalers. $2.54B in cash after Q1 fundraise. natural gas is the bridge fuel. nuclear is the destination. watch for a new entry here.

4/6. What I think he trimmed:

$BE — up 191% YTD. He’s disciplined about not holding consensus positions. partial trim is probable.

$SNDK — up 816% 1Y. Some profit-taking after that run is just basic portfolio hygiene.

$INTC calls — If these were exercised into direct shares during the Q1 rally, the position structure shifts. watch for reclassification from “call” to “SH” in the filing.

What I think he held firm:

$LITE — Nvidia locked this supply chain. you don’t sell a structural position after a procurement commitment that size.

$IREN $CORZ $CIFR — AI compute campuses. the thesis is still early. he’s not done here.

5/6. Three wildcard picks that fits his frameworks for the next AI bottleneck:

Wildcard 1: $CRDO

Build the cables and chips that move data between GPUs inside AI clusters.
revenue up 272% year over year. 88% market share in active electrical cables.
the plumbing of every large-scale AI cluster running today. market still treats it as a niche semi name.

Risk: copper faces long-term pressure from optical. customer concentration is real.

Wildcard 2: $ALAB

Builds the silicon that connects GPUs, CPUs, and memory so they can talk to each other at scale.

Q1 2026 revenue up 93%. CXL memory controllers shipping to hyperscalers now.

As agentic AI runs continuous multi-step workflows, memory-sharing becomes structural. astera built the controller for it first.

Risk: valuation leaves no margin for error. any hyperscaler moving in-house re-rates it hard.

Wildcard 3: $PENG

40 years of memory expertise. now building CXL-based servers that store AI agent context so inference doesn’t restart from scratch every step.

Integrated memory revenue up 53% YoY. AI business now 60% of sales. A tier 1 bank is already running their product.

Risk: Advanced computing revenue dropped 42%. Thin margins. Smaller operator in a space.

6/6. If you want to understand how this man thinks the playbook isn’t what he owns.

It’s WHERE THE BOTTLENECK MOVED.

I hold $IREN and $CIFR. Both appear in his book. That’s not a coincidence.

Which stock do you think Leopold Aschenbrenner is adding to next?

If you enjoyed this; share it, follow me, turn on notifications 🔔

I post my portfolio positions, deep dives, research and analysis.

100% free of charge. No subscription. Full transparency.

-BP

Note: This is not financial advice. Please research before investing.