Altimeter’s Gerstner: Scale wins superintelligence race as SpaceX plans $400B compute buildout
Brad Gerstner says we are on the brink of superintelligence and whoever controls the most compute wins, but the company doing the heaviest spending may not be the one pocketing the profits.
SpaceX (NASDAQ:SPCX | SPCX Price Prediction) closed at $159.31 on October 21, up 7.79%, after reporting third-quarter revenue of $14.60 billion. That same day, Altimeter Capital founder and CEO Brad Gerstner explained why he thinks Elon Musk is spending so heavily.
“Elon understands we’re at the brink of the singularity. We’re entering superintelligence. And whoever has the most scale, capital, compute and data wins the game of superintelligence,” Gerstner said.
Gerstner said Musk plans to stand up 10 GW of computing power next year. That would be about 25% of all U.S. capacity additions, and it would take roughly $400 billion of capital.
SpaceX now installs 200,000 GB300 chips a month, he added. A few months ago, that was a full year’s pace.
For you as an investor, the question is who keeps the money. SpaceX pays to build the capacity, while chip and memory suppliers are booking the profits today (we rounded up seven companies providing the AI expansion beyond the chipmakers in a free report).
Scale Thesis Drives a $2.10 Trillion Valuation
The logic is a loop. More computing power trains bigger models, bigger models draw more users and data, and that usage pays for the next round of computing power.
SpaceX’s results suggest the loop is working. Its AI Solutions & Infrastructure segment brought in $7.60 billion of revenue, up 830%, with help from the $60 billion Cursor acquisition. The loop breaks down if bigger models deliver smaller gains, if cheaper models catch up to frontier quality, or if paying demand grows more slowly than capacity.
Gerstner expects three gigawatts by year-end, and COLOSSUS II has already brought rated capacity to 1.7 GW. Getting to 10 GW only pays off if demand keeps up with each new gigawatt.
Capital Is the Bottleneck a Merger Would Ease
Few companies can pay $400 billion in cash flow. SpaceX ended the second quarter with $93.52 billion in cash and spent $15.83 billion on AI compute in that quarter alone.
Gerstner sees a deal with Tesla (NASDAQ:TSLA) mainly as a way to finance that spending. A combined company would have more collateral, so it could borrow more, at lower rates, and reach scale sooner.
His best guess for a Tesla merger is $520 to $550 a share. That is roughly 35% to 42% above its $386.47 close, even as the stock is down 14.06% this year.
The risk falls on Tesla shareholders, who would take on SpaceX’s spending even as Tesla’s own operating income fell to $718 million. A deal would also need approval from a majority of all outstanding Tesla shares.
Why Lenders Favor NVIDIA Hardware
Gerstner’s most original point is about collateral. A lender can value standard NVIDIA (NASDAQ:NVDA) GPUs more easily than custom chips, because many buyers could use a seized GPU.
That gives NVIDIA pricing power when credit is tight. A builder that borrows to fund an expansion will likely pay more for hardware a bank will lend against.
NVIDIA has built on that edge. It is working with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR to raise over $500 billion in outside capital, and SpaceX AI is a lead partner for its Vera CPUs. Companies that design their own chips may save on each unit, but they could end up paying more to borrow against them.
Suppliers Are Collecting the Margins Now
Gerstner cited NVIDIA’s 75% gross margin and Micron Technology (NASDAQ:MU) at 86%, both raised by shortages. He noted that NVIDIA stock has risen 18x in three years.
NVIDIA guided to roughly 74.0% gross margin, and management said it remains supply constrained. Those margins will likely narrow once supply catches up.
SpaceX carries the depreciation risk because GPUs lose value quickly. Cursor also added $1.2 billion of share-based compensation in the third quarter.
Key Risks SPCX Shareholders Face Next
SPCX carries significant risk at a $2.10 trillion market cap. Its shareholders pay for the expansion and take on depreciation, power, GPU supply, and execution risks tied to Musk’s timelines.
Circular financing adds risk because chip vendors are helping fund the customers that buy their chips. If demand stalls, that support turns into excess capacity.
NVIDIA appears better positioned. It sells chips to whichever lab wins, it has the financing edge Gerstner described, and its P/E of about 45 is far below Tesla’s 402.
NVIDIA closed at $221.97, up 22.82% over the past year, after making $59.69 billion last quarter. The next thing to watch is whether margins hold as Vera Rubin scales up.
Contact [email protected] for any questions or corrections.
Omor Ibne Ehsan is a writer at 24/7 Wall St. He is a self-taught investor with a focus on growth, cyclical, and dividend equities that have strong fundamentals, value, and long-term potential. He also has an interest in high-risk, high-reward investments such as penny stocks.
