A must listen. Already playing out with fears NVDA viewed as being more conservative on equity-funded neocloud buildouts. Ecosystem in the AI buildout will be impacted and stressed.
NEW ODD LOTS:
The problem with AI’s most bullish chart.
Before Anthropic starts trading, listen to me and @tracyalloway talk to @stanleylee21 and @pkedrosky about what’s really behind this chart of surging GPU rental prices. open.spotify.com/episode/4Nv…
25
The AI buildout will need to go from equity-funded compute purchases to revenue-funded compute purchases.
16
₿uzz YatyΞar retweeted
A more accurate title for the @wsj article would be:
“The completely legal backdoor that guarantees America will stay ahead of China in the AI race while giving America immense leverage in any future conflict.”
Legal under both the Biden and Trump admins. And these articles never mention the *American* hyperscalers that have GPU clouds in Southeast Asia.
Interesting share as always.
There is only one Jensen. The fear is real. $NVDA
There is real fear that Nvidia might pull back on their investments and backstops.
I believe this is misplaced. Nvidia’s competitors are eager to take their place, disrupt their relationships and fill any vacuum left by Nvidia.
There are at least three companies aggressively vying to be the new “Central Bank” of AI.
Across many @atreidesmgmt venture portfolio companies we are seeing Nvidia’s semiconductor and hyperscale competitors trying to lean in and displace Nvidia in important relationships. Ironically, sometimes at more favorable terms than those offered by Nvidia.
And of course, also beginning to see the first signs of Nvidia waking up to this risk and addressing it. Their strategic imperatives have not changed and their strategy was sound, but there is only one Jensen and competitors are now attempting to really take advantage of his absence. No one wants a monopsony at the lab layer.
I suspect the same thing is happening in the supply chain. But again, there is only one Jensen.
45
SpaceX’s orbital compute plans (Starmind / AI1 satellites) gain substantial strategic value if terrestrial power is the binding constraint on AI scaling, as Bill Gurley’s chart illustrates with China’s lead in nuclear reactor construction, multi-year build timelines for nuclear versus solar, and sharply lower costs in Asia versus the US and Europe.
Each AI1-class satellite is designed for roughly 120–175 kW average compute (peaking higher), powered by large solar arrays in sun-synchronous orbits that provide continuous sunlight without night-time gaps or weather losses, paired with deployable radiators that reject heat into vacuum far more efficiently than terrestrial chillers.
Musk has noted that 100 GW of space solar is roughly equivalent to 800 GW on Earth once batteries, capacity factors, and winter supplementation are accounted for, and SpaceX is targeting initial launches in 2027 with meaningful scale in 2028 while also pursuing large terrestrial clusters.
This architecture sidesteps grid interconnection queues, land use, and the slow nuclear buildout Gurley highlights, while leveraging SpaceX’s existing Starlink laser-link technology, manufacturing scale, and Starship launch cadence.
The result is a path to add compute capacity that is less gated by Earth’s power infrastructure, though it still depends on driving down launch costs, proving large-scale thermal management and radiation tolerance, and navigating orbital congestion rules.
$SPCX
38
₿uzz YatyΞar retweeted
If power is the real constraint for AI…maybe worrying about models and chips is a misplaced effort.
“NVIDIA has to become a frontier lab faster than frontier labs are becoming chip companies.”
Ok I do have a view on NVIDIA without Jensen.
I’ll tell you the real reason $NVDA will likely fail without him and why the frontier labs and hyperscalers are the biggest winners. (Elon has been in a really good mood recently. Have you noticed?)
The biggest threat to NVIDIA is its own customers, the frontier labs. NVIDIA has to become a frontier lab faster than frontier labs are becoming chip companies. And with Jensen out, they likely won't maintain the momentum to get there.
With recursive self-improvement (RSI), every frontier lab and hyperscaler must become a chip company or lose the AI race. It's that simple, and they've all laid the groundwork (Google TPUs, OpenAI and Meta building custom ASICs with Broadcom, Anthropic building an in-house silicon team, and SpaceX/xAI/Tesla building Terafab).
That's because software-only RSI has clear physical limits. Software can reduce how much computation and data movement a model needs, but it can't change the chip's physical bandwidth, memory capacity, or interconnect. Custom silicon gives labs another way to improve performance: designing the memory system, compute, and networking around their own workloads.
To get the real compounding effect of RSI, you have to co-design the model and the silicon so it becomes a two-way feedback loop: smarter models design better custom chips, and those chips train and run the next model faster. You can't do that on off-the-shelf NVIDIA GPUs because you get zero proprietary edge buying the same chip as everyone else, and getting your next-gen architecture into NVIDIA's silicon means sharing your roadmap with them years before tape-out (which they then sell to your rivals). Plus, AI now automates writing compilers and porting CUDA kernels to other chips, which kills the CUDA moat.
Jensen argued custom ASICs aren't more cost-efficient than GPUs because models change too fast for fixed hardware, and partners like Broadcom charge fat margins too. But hardware doesn't fall behind model changes when AI co-designs the next model and chip together instead of humans designing them separately (which Google and OpenAI are already doing). And labs aren't building custom silicon just to save on NVIDIA's 75%+ gross margins. Closing that hardware-software loop is the main goal, though the cost savings are still massive since RSI requires 24/7 internal inference where custom ASICs are way cheaper and more power-efficient than general-purpose GPUs.
That's why NVIDIA has been trying so hard to become a frontier open-weight lab, buying Groq (fast inference IP), Poolside + Nemotron (internal coding models), and Hugging Face (distribution). If closed labs win, enterprises just call their APIs running on the labs' own custom silicon and NVIDIA gets cut out. Subsidizing free open-weight models is the only way to make enterprises buy or rent NVIDIA GPUs to run the models themselves.
With Jensen gone, it's unclear if the Hugging Face deal even survives DOJ approval (can a GPU company own the whole stack?), and a non-founder is unlikely to take the massive risks needed to become a frontier lab: putting billions of capex into internal clusters and free open-weight models instead of just selling GPUs, while competing directly with its own biggest customers $GOOG, $META, $SPCX, $ANT, $MSFT
₿uzz YatyΞar retweeted
Many robotics and autonomous startups emerged because Jensen supported the community by sending dev kits to help bring up boards and integrate them into systems.
This never occurred in mobile phone systems, as most mobile SoC companies prioritize large enterprise customers who can afford expensive support fees, or phone companies could afford to make their own chips.
NVIDIA is different: it supports and incubates many small customers while they are still young, sending its best engineers to solve problems.
It's hard to imagine how this once-flourishing community will change if NVIDIA becomes like other corporations.
Wow cool
this might be the craziest thing humans have ever tried lol
a startup called Real Ice is literally trying to refreeze the Arctic
and on a small patch in northern Canada, it's actually working!
> in winter, they drill through the ice and pump seawater up on top
> the freezing air turns it into a fresh layer, so the ice goes into summer thicker
> their test patch came out about 50 cm thicker than the ice around it
this is huge because right now the Arctic is on track to lose almost all its summer ice as soon as the 2030s
and that ice helps keep the whole planet cool
it's like wearing a white t-shirt vs a black sweater on a hot day
right now the ice is the white t-shirt, bouncing the sun's heat back into space
once it melts, the dark ocean underneath becomes the black sweater, soaking all that heat up
and that extra heat speeds up global warming, which means rising seas, stronger hurricanes and deadlier heatwaves all around the planet
the company says thickening an area twice the size of California would cost about $5-6B a year
we're officially in the business of refreezing the planet
1
1
46
Yes
"X has become a much more powerful force in driving tech investor sentiment over the past few years. Honestly, besides Bloomberg, it’s my most important resource. I literally started a new section of my daily email a few weeks ago called SchilskyX where I screenshot and respond to Tweets. It’s that important. One important issue to note: the FinTwit community of influencers is VERY small. We’re talking less than 50 people that drive a preponderance of sentiment. Most investors and sell siders can’t tweet – so only a small number of FinTwit influencers with any sort of respectability drive much of the debate." - JPMorgan's Mark Schilsky
38
₿uzz YatyΞar retweeted
Replying to @CGasparino
Just normal shorts, Chuck.
Look, you are ugly and old and dangleously bejowled, but we agree more often than not. How about we stick to debating ideas?
As a great man once said: "I've never attacked him on his looks. And believe me, there's plenty of subject matter there."
Interesting
Delaware, we have a problem.
All ten NVIDIA directors who decided Jensen Huang’s fate are defendants in a lawsuit over how they handled him, and keeping him out helps them defend it, whether or not it is best for the company and its shareholders for him to be exiled. That is what we call a conflict of interest. Directors with a personal stake in the answer should not be the ones deciding it. The answer these directors have already given, full exile for Jensen, therefore deserves no deference.
The board must immediately call a special shareholder meeting, create new seats, and let shareholders nominate and elect independent directors to fill them, as NVIDIA’s charter and bylaws allow. Acting as a special committee with counsel of their own choosing, the new directors should receive the findings of the board’s investigation and decide afresh whether Jensen should return, in what role, and under what safeguards.
On November 11, a lawsuit named all of the outside directors as defendants, faulting the board, among other things, for keeping Jensen in charge while it investigated him. By the end of that day, Jensen was out.
As Reuters noted, Jensen’s removal stands to help the directors’ defense “by showing directors took action.” It answers the complaint’s charge that the board overly “deferred to Mr. Huang”: it creates evidence that they can say no to Jensen. Allowing Jensen to return, even in a safeguarded or circumscribed way, would undercut both personal benefits, even though it may be what is best for the company and its shareholders.
Any director deciding whether Jensen should return must ask what a return would mean for the company and its shareholders. These directors, however, must also ask what it would mean for them personally as defendants. Shareholders deserve a decision made by directors who have only the first question to answer. There is no shame in feeling the force of both questions; the directors are only human. The shame would lie in thinking they are above the law and refusing to step aside when they have such an obvious conflict.
Neither indemnification nor insurance removes the conflict. A declaration that the directors breached their duties to NVIDIA is among the relief the complaint seeks. An insurer can pay a judgment; it cannot erase that finding. As then–Vice Chancellor Strine put it: “No judgment in a fraud or other reputation-implicating case is cost-free.”
Delaware courts have accordingly refused to defer to directors whose decisions stand to shield them from adverse judgment for their own earlier conduct, and the case for recusal here is at least as strong. In Fitbit, the Court of Chancery explained that a director could not impartially consider whether the company should pursue claims against others when the company’s positions “would likely compromise his defenses” in a pending suit against him. The same is true here: reinstating Jensen is a position the company would take, and it would detract from these directors’ defenses.
In AmTrust, three directors facing a pending shareholder suit approved a buyout that would effectively end the claims against them. They had bargained for seven weeks and won the approval of two-thirds of the public shareholders. Yet the court still concluded that shareholders a material personal interest in the decision, that precludes deference. Here, far from taking their time and consulting shareholders, NVIDIA’s directors acted the same day they were sued and despite overwhelming evidence of shareholder disapproval.
Finally, in Revlon, the Delaware Supreme Court faulted a board for a decision that it said had, as a side effect, “to protect the directors against a perceived threat of personal liability” for their earlier choices. Revlon’s directors faced only a threat of suit. NVIDIA’s have already been sued. Thus, concerns of partiality are all the more salient here.
The directors may be right about Jensen. They may be wrong. What they cannot be is impartial. A decision this important deserves a process no one has reason to doubt. The board has failed to provide that. Jensen is not above the law. But neither is the board. The fix is simple: a special meeting, new independent directors chosen by the shareholders, and a fresh decision on Jensen from people who can make that decision without having to weigh both how it will affect the company and them personally.
1
92
😂 incredible.
Nvidia $NVDA leadership hosted a townhall meeting today in front of all their employees
Nvidia interim CEO Colette Kress reiterated her confidence in the company, “We’ve got this,” she told staff
The company made it clear that co-founder and former CEO Jensen Huang won't be coming bac
Kress called Nvidia employees "brilliant minds" and said the mission would remain the same, with the company seeking to rebound from upheaval and scandal.
The meeting ended with the Tom Petty and The Heartbreakers’ song “I Won’t Back Down.”
(Source Bloomberg)
46
It was always this. Prepare for this risk to reveal itself.
Talent finally has the out to hang up the coat. The exodus of Nvidia talent has already begun.
I'm hearing multiple things from industry executives:
1. Potential $NVDA employee exodus if Jensen doesn't come back soon.
Many senior management and engineers can easily retire after Nvidia stock appreciation over the years.
They don't need to continue working sitting on millions in equity. But they continued to work because Jensen was a visionary leader.
Under new interim CFO (former CFO), it gives an excuse for that lot of people to finally quit soon. A pay bump or hiring $ARM CEO isn't going to change that.
So you have generational talent leaving.
2. Loss of confidence in $NVDA future growth (Jensen was the moat).
What I'm hearing is that Jensen understands heavily top-down architectures, and it's this magical discretionary element that doesn't get covered in media.
When cuda, network, and $NVDA internal teams are arguing, Jensen made the final choices around heavily nuanced discretionary decision about where Nvidia + AI ecosystem ends up going:
Then everyone followed. The supply chain, capital, agreements, and faith.
You also can't suddenly replace the trust that local Taiwanese factories build with Jensen over the years if you bring in external CEOs, or even internal executives.
It was transformational bet after bet after bet after bet that Jensen made until Nvidia became a former $5T+ company.
3. $NVDA is not that technologically far ahead of others. It's not a 5 year lead, but months ahead to a year in different areas.
Jensen was accelerating all of that. Longer Jensen is gone, the longer Nvidia and its competitive ecosystem decays, beyond the current near-term roadmaps.
Nvidia's greatest moat was Jensen Huang making high-stake architectural decisions and accelerating capital, suppliers, + the industry to rally around his vision.
There's a massive loss of confidence in $NVDA with Jensen gone.
40
Interesting
We shrank 1 GB of Wikipedia to 105.39 MB, losslessly.
Introducing tufazip: the best entry on the Large Text Compression Benchmark using a neural network as its sole predictor. Everything above it mixes a transformer with hand-built models.
Full post 👇
tufalabs.ai/research/tufazip
28

























