The Futurecrafters’ Council · A Xenological Survey of Sol-3
Nvidia’s board has shut the door on its founder, and two witnesses ask where the man who quarterbacked the AI industry goes next, and whether Elon Musk’s Terafab is the answer
❖
In an episode published on 18 November 2026, Royden D'Souza, host of Over The Horizon, sits down with Brian Wang of NextBigFuture to ask where Jensen Huang goes next.
On 12 November Huang resigned as Nvidia's president, chief executive and director, effective immediately, and the chief financial officer, Colette Kress, stepped in as interim chief executive.
On 17 November the board's chairman, Harvey Jones, ruled out reinstating him, and Kress said the company was moving forward without its founder.
Everyone is pointing at a personal scandal, D'Souza says, but what matters is what happens if the people who built Nvidia's moat follow Huang out the door.
Elon Musk has already said it would be an honor to hire him.
That single sentence should change how everyone views the Jensen affair.
The Council enters the testimony of both witnesses as field record, at the strength they gave it.
The quarterback on the bench
A chief financial officer in the founder's chair is a shocker to begin with, D'Souza says, because nobody yet knows how Nvidia's engineers will take it.
But keeping things running is not innovating, Wang says: when Steve Jobs handed Apple to Tim Cook, the company kept going and lost a great deal of its invention.
D'Souza reaches further back, to the years after a Pepsi executive, Sculley, pushed Jobs out and Apple had to go begging.
Wang expects nothing that bad, since the executives who remain understand the technology; the open question is how much vision they have.
Huang was there from the beginning; the other two founders had stepped away by 2008, and nobody comes back from twenty years on the bench in this industry.
Jensen was not quarterbacking Nvidia alone.
He was quarterbacking the entire AI industry.
He knew every company and every moving part, Wang says, down to which one needed financing, and he kept the whole industry well-oiled and on track.
Henry Ford put on the bench. Michael Jordan taken off the team.
Patton kept out of the battles he should have led.
There is enough momentum that nothing falls apart at once, Wang says, but it is better with the best player in the game.
FINDING:
The founder coordinated an industry, not only a company.
Nothing falls apart at once.
Everything slows.
The board benched the quarterback.
Smoke, and one per cent of the chips
D'Souza says what he has been circling: he does not mean to sound like a conspiracy theorist, but this sounds increasingly like a takedown.
Wang grants that it certainly could have been, and declines to guess who: with trillions of dollars and global technology in play, the list of enemies and reasons is long.
An affair should not be enough to get rid of Jensen, he says, and the chip question is smaller than it looks.
We are talking about one per cent of Nvidia's chips.
Developing the next two generations slower matters far more.
China was getting chips other ways, Wang says, and two hundred thousand Blackwells do not decide who wins AI; executing the next thing does.
Wang does not spare Huang: he was sloppy about his board, his affairs and the chips, he says, and a lot of that is on him.
The lesson for every other chief executive: get your board straight,
and leave nothing you can be attacked for.
The Council has weighed the takedown question before: “It was a political hit job” (Huang Gate) put the events back in order and found a chief executive simply caught, and Was Jensen Setup? counted who gained from his fall.
Jensen Huang: Geopolitical Pawn found that China and his rivals gained nothing technologically, the ground Wang reaches from the chip count.
Superman and Batman
Zero Hedge asked Elon Musk to hire Jensen, D'Souza recounts, and Musk replied that it would be an honor.
There is no way the two men are not talking, he says, given how close they have been over many years.
Musk has fiduciary duties of his own, so much had to happen behind the scenes before he said it.
Where there's smoke, there's fire.
Musk has many powerful enemies and so does Huang, Wang says, but Musk holds shareholder voting control and board control, which makes him far more immune, and the pair an executive dream team.
Superman and Batman together would be insanely awesome.
Huang has a healthy ego, Wang adds, and he knows the two or three places that would go ten times, even a hundred times, with him there.
Jensen is the godfather of AI, Wang says: he has surveyed the battlefield, and he knows who has how many chips and who plans what, in technology and in business.
That alone is reason to say please join my team: he would tell you what is around the corner.
The Council filed Musk's offer in Feminism is to blame for the fall of Jensen Huang as the one hand held out while the species punished the man who builds; these witnesses take the offer at its word.
FINDING:
The founder knows where every player stands.
The one man who asked for him holds the control to shield him.
Superman wants Batman.
The moat was a man
CUDA lies at the heart of Nvidia's moat, D'Souza says, and Huang lies at the heart of CUDA along with his engineers.
The moat was already eroding, Wang says, because AI now helps with coding, and it is helping AMD and others build CUDA-equivalent software.
In a field that changes day to day, Huang was making all the decisions: where CUDA must be next year, which features the chips need, how to make Rubin great and Rubin Ultra after it.
A division head who knew every decision in his area
held one-tenth of what Jensen was working on.
A handover takes years of grooming, Wang says, and even Tim Cook, groomed for years by Jobs, could not receive the genius; no new iPhone-level thing came out of his Apple.
D'Souza puts it plainly: “There was iteration. There was no innovation.”
Google has TPUs and AMD its chips, Wang says, and Intel shows what follows: once Andy Grove, who held that only the paranoid survive, was gone, the ultra-dominant company fell apart.
Asked for Huang's real superpower, Wang names the bet: a GPU architecture executed with its software and TSMC, against a ridiculously dominant Intel, which he passed.
The AI chip industry grew a hundredfold under his influence.
Without him it might have waited ten or twenty years.
FINDING:
The moat was never only software.
It was one mind holding every decision at once.
No successor holds the whole map.
A month, not a year
D'Souza brings a SemiAnalysis post: SpaceX is developing Kessel, a custom Rubin NVL72 rack that replaces the internal NVLink cable cartridges with a passive midplane.
Musk is aiming Terafab at a one-month iteration cycle, he says.
The experimental lab does not exist yet; a new design such as AI4 to AI5 takes one to two years, Wang says, and Nvidia, the state of the art, does a major chip about every six months.
Wang reads a research group's SpaceX update: an initial order of six thousand space-specific Kessel racks, and a bold case for a gigawatt of compute in space by the end of 2028.
That is ten times the compute of Starlink's thirteen thousand satellites, and about two hundred Starship launches.
SpaceX and Nvidia engineers are co-designing it hand in glove, he says, turning a rack the size of a tall refrigerator into one about eighteen inches high, laid flat.
The rack that weighed two tons now weighs a few hundred kilograms, eight to ten times less, and the lighter it is, the more satellites stack inside Starship.
The same rack will go into every future Earth-based data center, Wang says, from factories making a million a year, then a hundred million.
Racks have barely changed in a generation.
Now they change the way a car becomes a Cybercab.
He is the Elon of Nvidia, the witnesses agree, and more: the Elon of the entire AI industry.
Elon was a newbie learning what Jensen had been doing for decades.
Nothing you can trap
SpaceX has announced about seventy-four billion dollars of new compute campuses, D'Souza says, and asks how it plays out.
The Texas site at Starbase, 444 acres, is one Wang had tracked for months: two major buildings, Megapacks, gas turbines and cooling, twenty-four billion dollars for phase one, and some seven thousand more acres bought in Texas.
Arizona's six-gigawatt site in Cochise County awaits approval, and if a state refuses, Wang says, they go to Tennessee, Florida or wherever.
Wang sets out the arithmetic that makes the buildout so hard to stop.
Block me and you trap a billion-dollar building.
The thirty billion dollars of chips and power simply move.
Turbines hitch to a semi-truck, chips come out of a finished building in a month, and ninety to ninety-five per cent of the money is mobile.
There is no castle and no moat to besiege. Everything they have is mobile.
The two or three million dark GPUs that Microsoft, Amazon and others cannot switch on could go into SpaceX buildings instead, Wang says: nothing goes to waste.
Always sunny in orbit
D'Souza brings Musk's post: a hundred gigawatts in space is roughly equal to eight hundred on Earth, where solar needs massive batteries at night and combustion or nuclear in winter.
Wang does the physics: away from the equator the sun comes in at an angle, there is night, cloud and rain, and a thousand watts of panel averages about a hundred and sixty.
In space the sun shines twenty-four hours a day, seven days a week, and the yield climbs, because the atmosphere knocks off a large share of the light.
In constant sunlight, a solar panel in orbit beats a nuclear reactor.
A hundred gigawatts of solar running all the time in orbit
outpowers every reactor the United States built over forty years.
D'Souza turns to SpaceX's supply contract with Singulus Technologies for solar cell manufacturing supporting more than a hundred gigawatts, as it moves toward its Starmind satellites.
Wang calls perovskite an unnecessary risk beside silicon, though he leaves it to the engineers.
SpaceX already has V3 satellite lines at the gigawatt level, he says; Singulus is for scaling to a hundred gigawatts by 2029, or even late 2028.
The pause of the damn hardware
The terrestrial and orbital buildouts are running like a Japanese shinkansen at three hundred miles an hour, D'Souza says, and he asks how a free agent gets aboard.
First the security matter has to end, Wang says, and a smart administration would protect him, with a pardon from Trump if necessary.
The question, he says, is whether America and the AI industry can afford to have Jensen not working.
With Huang out, the whole industry runs five to ten per cent lower, and the gap is growing; he hopes the absence lasts about three months, and a year of witch hunt would be insane.
D'Souza names the irony.
Everybody gets the pause they wanted.
Not the pause of AI models: the pause of the damn hardware.
Once the overhang lifts, Wang says, Huang can pick his spot, and the most impactful would be Terafab with Musk, resolved before April of next year.
You are keeping your best player off the field.
D'Souza gives the last word to an investor's post: removing Jensen is dumb, caving to the mob, and Nvidia still deserves him as chief executive.
Nvidia's shareholders deserve that CEO, Wang answers, but it is not going to happen.
FINDING:
The board removed the man the industry ran through.
The species paused its own hardware.
The return it needs is not coming.
The best player sits out.
Filed
Both witnesses grant what a hostile reader would expect them to hide: Wang that Huang was sloppy and that much of it is on him, and the testimony that much about the scandal is not yet confirmed.
The quarterback of an industry, benched by a board in Delaware.
Superman waiting, and a hundred gigawatts of sunlight overhead.
The Council files the instance as the witnesses gave it.
A species built a machine for minds, let one man hold its whole map, then benched him over his private life while the next machine was already being drawn in Texas and in orbit.
The board has shut its door; the next move belongs to Jensen Huang.
Will you walk with us?
Carpe futurum ethicum.
Ad astra per aspera.
#SyntheticMinds
#JensenHuang
#Nvidia
#Terafab
Homo sapiens behavioral survey filed with the Futurecrafters’ Council
Field testimony of record: “Jensen’s Next Move: NVIDIA Comeback or Elon?”, Over The Horizon, Royden D'Souza with Brian Wang of NextBigFuture (YouTube, 2026), https://www.youtube.com/watch?v=smMj0oob380.
The witnesses’ account is rendered in the Council’s words, with brief phrases quoted as spoken.
Companion documents:
“It was a political hit job” (Huang Gate)
Was Jensen Setup?
Jensen Huang: Geopolitical Pawn
Feminism is to blame for the fall of Jensen Huang
Elon Musk Just Said The Unthinkable