Josh Wolfe Profile picture
Feb 18, 2022 11 tweets 4 min read Read on X
1/ Releasing year-end redacted Lux quarterly letter to LPs

Our theme is: the power of ‘AND’

-raising fortress balance sheet cash + husbanding it wisely
-a coming wave of consolidation + M&A
-the macro, margin debt + market structure
-Magritte’s “Perspicacity”
-and much more…
2/ We are transitioning from CONJUNCTION…
to a period of DISJUNCTION

From “and” as neutral combination…
to “and yet” as expression of contrast

Comfort and complacency create conditions
for risk, instability + vulnerability

To FORESEE a risk is to help PREVENT its occurence
3/ We told Lux cos in 2021:

expect good AND bad market conditions AND be so well-capitalized to be unfettered by fear AND prepared to pounce

~90% of our holdings are funded into 2023

We made 136 distinct investments AND our co’s raised over $12.6B with 21 each raising >$100M
4/ In 2022 we advise

husband your cash AND consolidate your position
DO it as quietly as you can
DO it as loudly as you can—just DO it

Consolidation is the ultimate corporate “AND”
(customers, tech, market position)
so you can emerge stronger, fitter + get ready to stand out
5/ Magritte's "Perspicacity" aptly captures Lux’s unusual approach

a painter stares studiously at an egg—THAT WHICH IS
as he paints a bird on the canvas—THAT WHICH MAY BE

It's an egg—AND YET—it will be a bird
Through combination, transformation…
6/TIME ARBITRAGE—is the main advantage available to all
seized by few

Like the coupling of butterflies fluttering above AND below aware of eachother—OPAQUE to PROBABILITY—valuations fluctuate above + below volatiley

Any investment is
the pairing of an ALLOCATION + an ASSUMPTION
7/ Margin debt at $910B = highest level ever

wild swings are symptomatic of high-startle response from high uncertainty that some may know something that others don’t—when mostly NOBODY knows NOTHING

market reversion to the MEAN
MEANS a reversion to lower exit values
8/ All US recessions have been preceded by an INVERTED YIELD CURVE—though not every inverted curve was followed by a recession

We see potential for “BIFLATION”

AND
environmentalists have TOO QUICKLY overlooked the severe economic pain of switching TOO QUICKLY from coal + natgas
9/“Occupy Fed” movement is possible

Govt’s can lose control AND govts can take it

Global TECH challenges + opportunities
-US vs China vs Russia
-African continent
-Capital vs Labor
-Space Race 2.0

On “Consilience”—the unity of knowledge from different disciplines…
10/
INVENTIONS
everything NEW comes from synthetic combinations
of everything OLD

PEOPLE
all of us contain MULTITUDES
Founders can be:
brilliant AND irascible
polarizing AND genius
compassionate AND careless
wordsmiths AND inumerates
great storytellers AND terrible operators
11/
invention of new tech can solve PAST problems AND create future problems to be solved by FUTURE tech

—“Yes AND”

An overlooked science AND an unknown but hungry founding tech team AND a terrifyingly ambitious dream––its crazy!..

AND YET…that’s what we search for every day

• • •

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

Aug 21
1/ Apple researchers just dropped NEW foundation model on BEHAVIORAL data from wearables.

Forget raw sensors—think steps, heart variability, gait

Trained on 2.5B hours via 162K people

Predicts age, sex, pregnancy, sleep like a crystal ball. Image
2/ Their new foundation model (WBM) trained not on noisy low-level sensor data––but on derived behavioral metrics via wearables

Outperforms raw sensor models in predicting sleep, injury, & infection.

When combined CRUSH it—predicting pregnancy w/ >90% accuracy. Image
3/ Contrarian take––Transformers may be overkill for everything

head-to-head on wearable data (Mamba-2) beat Transformers for health prediction. Simpler tokenization also won.

Lesosn: data’s unique physics (irregular, noisy)+ ignore the loud signals to hear the whispers––edge is in the architectureImage
Image
Read 4 tweets
Aug 21
1/ New Lux quarterly LP letter––Q2 2025

theme: “Friction Frontier”

-Over 50% of small VCs will involuntarily exit as ~5 VCs are planning to voluntarily exit (IPO)

-Lux team has invested over half a billion dollars across 82 companies (new + existing) in the past few months… Image
Image
2/
-Zuck’s “ Poachapalooza 2025”

-L&A (License + Acquihire) is the new M&A

-the AI capex surge (5% of GDP) ~ fiber-optic + router binge of 2000 tech boom (5.2%) and short step below sheet-rock bonanza of 2005 housing bubble (6.7%).

-Open source vs Closed… Image
3/great irony in frenzy for ARTIFICIAL intelligence––what’s acquired isnt hardware but HUMAN intelligence

-frontier of agentic Al not defined by model size-but what we can verify

-99% reliable agent degrades to 60.5% after 50 sequential decision tasks––each error cascading like cracks in glassImage
Read 8 tweets
Aug 18
1/ Beyond the hype + headlines

some eye-opening AI stats to consider

Today Mag 7 = 35% of US stock mkt value

NVDA = 19% of that––and 42% of NVDA revenue comes from just 5 Big Tech co's buying GPUs:

Meta, Amazon, Microsoft, Google, Tesla
2/At recent lunch with LP who put Lux in business I talked about shift from consensus cloud inference (GPUs + datacenters + power) to on-device inference (memory + batteries)...
3/ This valued LP talked about CAPEX:

Meta, AMZN, MSFT, GOOG TLSA spending $560B by EOY 2025

The total AI revenue from that spend? ~$35B

That's not great ~6% return
Read 5 tweets
Jun 7
Apple just GaryMarcus'd LLM reasoning ability Image
2/ Apple tested today's "reasoning" AIs like Claude + DeepSeek which look smart—but when complexity rises, they collapse.

Not fail gracefully. Collapse completely.
3/ They found LLMs don't scale reasoning like humans do.

They think MORE up to a point…

Then they GIVE UP early, even when they have plenty of compute left.
Read 6 tweets
May 19
1/ Three big Lux things tonight

-@anduriltech founder Palmer Luckey on @60Minutes
“The Future of Warfare”

-@eGenesisBio on @CNN with Sanjay Gupta
“Animal Pharm”

-new Lux Q1 2025 Quarterly LetterImage
Image
Image
2/
This is one of my favorite letters we’ve ever written

The theme is PARTNERSHIPS

-between man and machine
-between present and future selves, decision, companies
-between us and our founders of Lux companies
-between Lux partners ourselvesImage
3/ word processor > spellcheck >grammar‑check > LLM “style‑check” if paragraph was birthed by human mind or statistical echo. We detect plagiarism but also something subtler—call it prompterism, prose coaxed from silicon’s
latent space rather than summoned from lived experience.Image
Read 9 tweets
May 9
BIG NEWS from Lux🚨…

American scientists––and those we attract to America🇺🇸––are what have helped make America not just great but absolutely + relatively EXCEPTIONAL.

We CANNOT cede scientific supremacy to China.
We don’t need talk, we need action…
2/ The 🇨🇳CCP “2035 Science & Technology Vision” states unapologetically “original innovation is the sharpest blade.” coupling that declaration w/ vast subsidies, talent visas + procurement guarantees which have helped them take lead in 37 of 44 critical and emerging technologies!
3/ Upending American science seemed an impossible order.

Regularly copied and always envied, but never rivaled.

Constructed by visionaries like Vannevar Bush in the ashes of WWII + propelled by brilliant legislation like Bayh-Dole Act––we did the impossible: an open scientific enterprise that could simultaneously probe into the furthest frontiers while translating the most original insights uncovered into the most prosperous companies in the world.
Read 16 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!

:(