Josh Wolfe Profile picture
Nov 23, 2024 17 tweets 7 min read Read on X
1/ New Lux quarterly letter to LPs...

Our theme? Tech, venture, global markets told in...

...Four parables of Four brothers
1 Prometheus
2 Menoetius
3 Epimetheus
4 AtlasImage
2/ The The Prometheus Promise––is Extending Human Limits what we call extensionalism

technology is not just augmenting our natural senses, but fundamentally extending and transforming them inImage
Image
3/ From sight, sound, touch, hearing and now…smell.

Technology extends our senses (“extensionalism”) with many Lux family co’ leading the way from

@physical_int + CTRL-Labs + Auris Surgical Robotics (touch)
@EikonTX + @Matterport (sight)
Chromatic (hearing)
@Osmo_Labs (smell)Image
4/The lever + pulley of antiquity gave way to the steam shovel + excavator

Even the modern state's quest for legibility––what James Scott (Seeing Like a State) saw as drive to make complex systems measurable + manageable––Lux cos extend sensed in sea + space @saildrone @planetImage
@physical_int @EikonTX @Matterport @Osmo_Labs @saildrone @planet 5/ From genetically modified pigs with life-saving organs successfully transplanted into humans in dire need with no alternatives…

…to AI infrastructure with important 2nd order effects underway…. Image
@physical_int @EikonTX @Matterport @Osmo_Labs @saildrone @planet 6/
Today's largest Al clusters use around 100,000 Nvidia H100 chips and consume roughly 100 megawatts of power. Next year's clusters will scale to 300,000-500,000 chips requiring nearly a gigawatt of power — about the consumption of a small city. Image
7/ we see 2 futures emerging

1 Distributed compute can eliminate need for massive monolithic clusters––as per our investment in @togethercompute

2 As industry fixates on bigger data centers + more powerful chips––Apple quietly has showing how to run models directly on devices w/ limited memoryImage
8/ America only has a handful of locations that can provide the gigawatt-scale power these massive clusters will require.

While we're bullish on the seeming resurgence of elemental energy - our rebrand of nuclear––abundant natural gas from the Texas Permian seems a wiser bet.Image
@physical_int @EikonTX @Matterport @Osmo_Labs @saildrone @planet @togethercompute 9/ NYC's emergence as a major Al hub with 20 new unicorns in 2023 and 10 more already in 2024.

BUT…beneath the surface excitement though, we identified critical insights that are grounding our investment approach from the irrationality of the markets Image
10/ Epimetheus represents the danger of divorcing motivation from consequence. The benefits of the future start with tradeoffs of the present constrained by decisions of the past.

Higher for longer is our posture on the natural cost of capital, regardless of what the Fed does.Image
@physical_int @EikonTX @Matterport @Osmo_Labs @saildrone @planet @togethercompute 11/

We have 4 main strategies for current market:

1 Lux Labs
2 Corporate Spin-offs (globally)
3 Tactical Global Opportunities
4 “Fixware” (aka maintenance) (globally) Image
12/ While techno-optimists focus on frothy greenshoots, they're missing the industry's further desertification: over 30% of 50,000 private cos are trapped in late-stage limbo while >730 unicorns are trapped in exit-less purgatory, with nearly 50% over 10yrs old + dragging down returns like sea-sinking anchors…Image
13/ We are particularly excited about Lux Labs

14/ Lux Labs formalizes our nearly twenty-year track record of transitioning breakthrough science from the laboratory to the marketplace.

We have successfully spun out or have been first investors behind de-risked technologies and teams from big tech. Examples include Osmo and Physical Intelligence from Google, Aeva (and another not yet disclosed) from Apple, and Evolutionary Scale from Meta. These unlock value by giving breakthrough teams and technologies the funds, freedom and focused entrepreneurial attention to thrive.Image
@physical_int @EikonTX @Matterport @Osmo_Labs @saildrone @planet @togethercompute 15/
Globally we are particularly excited about Israel + Japan +UK
🇮🇱🇯🇵🇬🇧 Image
@physical_int @EikonTX @Matterport @Osmo_Labs @saildrone @planet @togethercompute 16/ Image
17/ As we look to the remainder of 2024 and beyond, several trends bear watching:

the acceleration of Al capabilities beyond language models into biology, chemistry and physical systems; the growing imperative for domestic manufacturing and supply-chain resilience; the continued convergence of commercial and defense technology; the evolution of experimental science through automation; and finally, the transformation of healthcare through sci-fi breakthroughs like xenotransplantation.

In the end, we return to where we started: compassionately competitive, motivated and mission-driven, helping to write the future, one breakthrough at a time.

We're betting on the impatient optimists who refuse to wait for permission to build the future.Image

• • •

Missing some Tweet in this thread? You can try to force a refresh
 

Keep Current with Josh Wolfe

Josh Wolfe Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @wolfejosh

May 17
1/ Preview of a speculative hypothesis

hormuz > fertilizer shortage > vulnerable food supply chain > food insecurity > existing migratory pressures exacerbated > peripheral country crisis in EU
=mkts 📉

unassimilated immigrants + rise of Right > violent clashes in EU
=mkts 📉 Image
2/ Complex systems (via @sfiscience) fail not from obvious blows

but from seemingly inconsequential events (a "single spark") igniting volatile _ pre-existing conditions.

endless historical precedents of small spark collrelations: ex #1 Russian Wheat > Arab Spring... Image
@sfiscience 3/ Ex 2: single shot fired in Sarajevo > WWI Image
Read 6 tweets
May 8
1/ On Feb 21 I told my boss aka my wife @LTwolfe that I had my third conviction call not yet appreciated + to buy a basket

*In 2016 it was NVDA, pitched publicly at Invest For Kids in Chicago (80x+) when Lux portfolio company Zoox was using NVDA chips and I predicted the narrative would change from gaming consoles PS4/Xbox to simulation and AI…
2/ here was the 2016 tweet on NVDA…
3/ In the fall is 2024 I pitched SK Hynix, Micron, Samsung and we bought in our pa

Mostly because I had read a technical paper about LLMs and inference on device with local flash (again changing narrative from commodity memory players to coveted attach rates on HBM AND edge)
Read 10 tweets
Mar 18
1/ New paper from @ylecun et al on alternative approach for AI to learn more biologically...

paper basically says AI is super smart but still can't learn like a toddler can...

the main critique Image
the main critique...

is that top models today dont learn once deployed.

and curating data + training pipelines + filtering out noise is all done by humans in loop instead of being an intrinsic skill of the AIs...

here's Claude explaining the paper simply Image
3/ not quite Kahneman style but they call it System A +B

System A = learning from observation

this is self-supervised learning like predicting next word or video frame. good at making representations but passive and struggles to understand cause + effect... Image
Read 7 tweets
Feb 21
1/ The shift from GPU capex (NVDA) to memory intelligence is already underway (SK, Samsung, Micron).

What will be next?

Even the best HBM chips are constrained by how they get architected into systems...
2/ Nvidia's GPU racks have a fixed compute-to-memory ratio. It is baked in + hard to change.

It is GREAT for training––and a WALL for memory-intensive inference...
3/ We found a company with 1,000x more memory than a standard enterprise server.

1 of their servers replaces up to 10 conventional racks.

Less floor space. Less power. Less cooling. Same or better output...
Read 7 tweets
Feb 20
New Lux LP letter dropping this weekend...
2/ Q4 2025 Lux LP Letter… Image
3/ the DILUTION DELUSION
vs CONFIDENT CONCENTRATION

physics teaches that systems in disequilibrium eventually corrrect and the questions is never whether, but how––and how violently.

excess energy must dissipate whethery by gradual cooling or explosive release… Image
Read 11 tweets
Dec 5, 2025
1/Today at private investor gathering I pitched

LONG
Adobe (ADBE)

SHORT
basket of scientific publishers
Wiley
Elsevier
Springer Nature

AI overestimation oversold Adobe (down 40%)

AI underestimation along with 2 other catalysts is set to crack the oligopoly of publishers
2/ first $ADBE thesis then the short basket + weighting

reasonably clean dislocation in large-cap SW

Stock ~$325 (down ~40% from highs)

Mult compressed from ~40x to ~20x

EV/EBITDA ~11–12x (10yr low)

mkt sees "value trap" facing existential AI threat

I see category error
3/ I'll argue the consensus bear case:

"AI kills the seat" + commoditizes design

one AI-augmented designer replaces three

Adobe loses its moat, got disrupted, growth is slowing + it’s milking a hostage customer base...
Read 25 tweets

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Don't want to be a Premium member but still want to support us?

Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal

Or Donate anonymously using crypto!

Ethereum

0xfe58350B80634f60Fa6Dc149a72b4DFbc17D341E copy

Bitcoin

3ATGMxNzCUFzxpMCHL5sWSt4DVtS8UqXpi copy

Thank you for your support!

Follow Us!

:(