Twitxy Saved

Brett Winton

Verified account ARK Invest

@wintonARK

Chief Futurist @ARKInvest. ARK Venture IC. Welcome to the Great Acceleration. ark-invest.com/terms/
Venice, CA ark-invest.com Joined August 2012
601 Following 217K Followers 13K Posts

Latest posts 5

Brett Winton Verified account @wintonARK
relative to our (aggressive) expectations

Starship landed on its 2nd commercial flight (we were expecting 6th at the midpoint)

Seems to be refurbishable within weeks (we were expecting months at the midpoint)

And is actually truly reuseable (we thought there was a ~5% chance they would never solve re-useability with a heat-shield approach)

At root, SpaceXSI is growth-governed by asset turns.

Relative to the non-reuseable case, SpaceXSI with a re-useable Starship stack becomes >2x as valuable by 2031 (per our research)

And the improved economics suggested by the faster rate of progress on the rocket itself suggests a ~10% increase in 2031 value and revenue relative to our expectations.

All of this to help explain why the stock is

[checks notes]

~flat since they caught the largest ever object to return from space in the grips of a giant steel set of chopsticks.
14 replies 15 reposts 230 likes 8.5K views View on X
Brett Winton Verified account @wintonARK
The market's $70 trillion mistake

By basing frontier lab topline trajectory on revenue accrual, software analysts are pulling a tool out of their toolbox that does not apply, ignoring history, and potentially wildly underforecasting frontier lab trajectory.

This feeds into systematic mispricing throughout the AI value chain and is (potentially) the root cause of the market's broader skepticism for AI.

Here I make the case for AI on the merits of what the companies have built and are building and show how the compute-based approach yields an expectation that differs from the market's by trillions.

(As can be seen in the charts, dotted line showing the market's approach, purple showing what we believe to be the correct approach).

will explain more in the thread
👇
5 replies 7 reposts 57 likes 9.7K views View on X
Brett Winton Verified account @wintonARK
in a $40 trillion market with 3 and perhaps 4 players, that SpaceX can provide the key supply input at the margin at scale (and can alternatively use that supply to directly compete in the market), results in an equilibrium where the marginal price per compute can get to obscene levels.

the simple logic is that openAI and anthropic have non SpaceX future compute reserved, and so enter into a race condition to train and serve a somewhat sticky market so that they can maximize internal margins on the reserved compute once the market has saturated; even knowing that SpaceX will eventually pull compute in-house to occupy the majority of the market this is optimal strategy.

Even knowing that they won't be able to occupy the majority of the market due to spaceX's compute dominance, they overpay in the near term to maximize their long-term profitability serving a meaningful minority of the market with the compute they do have secured.
12 replies 18 reposts 193 likes 13K views View on X
Brett Winton Verified account @wintonARK
the merest beginning
1:24
Brett Winton Verified account @wintonARK 3d
This is the largest ever object that we have safely returned from orbit

Drops cost per kg in orbit to roughly $100 (from $1000) and promises a massive increase in achievable annual upmass.

With this cost input falling out of the equation the cheapest electrons in the solar system will be harnessed in orbit rather than on earth.

That this has happened as we have discovered mechanisms by which to directly transfigure inexpensive electrons into digitally transmittable machine intelligence that can directly accelerate economic production, is either:
a happy accident;
a fairly convincing proof that we live in a simulation (built atop a random number generator seeded with the number 42);
or a testament to Elon Musk’s ability to synchronize and accelerate technological and engineering breakthroughs, piecing together convergent technologies to bring forward the singularity.

(Or perhaps all three.)

Where once we were limited by the resources we could extract from the ground, we now have access to the energy-mass of space and the seeds of the tools that we will need to begin to order and arrange it.

A moment. A historical mind blowing moment. And the merest beginnings of the beginning.
View quoted post on X