Sean White @SeanWhite_NTSB - Twitter Profile

Sean White

@SeanWhite_NTSB

3 minutes ago

Exactly

Steffan Szumowski

@UnoMasReactor

about 3 hours ago

Anti-nuclear activists don’t want “long term waste solutions”. They want used nuclear fuel to be perceived as an unsolvable problem… Is there something terrifying going on in this picture that I’m missing?

UnoMasReactor's tweet photo. Anti-nuclear activists don’t want “long term waste solutions”. They want used nuclear fuel to be perceived as an unsolvable problem…

Is there something terrifying going on in this picture that I’m missing? https://t.co/ZHHyg5u8BT

5

20

1

0

1K

0

1

Sean White

@SeanWhite_NTSB

1 day ago

An epic and wonderful novel - all Americans should read this fabulous story

Packy McCormick

@packyM

1 day ago

I looked into why everyone is reading Lonesome Dove. https://t.co/UGc5kPVmJe

34

201

6

191

151K

0

38

SeanWhite_NTSB retweeted

Packy McCormick

@packyM

1 day ago

I looked into why everyone is reading Lonesome Dove. https://t.co/UGc5kPVmJe

34

201

6

191

151K

SeanWhite_NTSB retweeted

David Sacks

@DavidSacks

1 day ago

Narrative violation: According to the Economist, AI has created 1 million new jobs in the U.S.

261

7K

990

1K

652K

Who to follow

Chief Economist @Huntington_Bank. Former lead forecaster @BLS_Gov and chief economist @atlchamber

S. Mohsen Fatemi

@MohsenFatemiii

PhD Candidate @KUSPAA | Co-Director Secretariat @PMRA1991 | Researching Energy Governance, Policy & Justice | Sustainability Adv. Board @LawrenceKS

Sean White

@SeanWhite_NTSB

5 days ago

A revealing window into this growing and very large part of the US economy. 🤔 @EricKno67379868 @stbeden

Arvind Srinivas

@glocalinvestor

12 days ago

Morgan Stanley issued a very interesting report on the AI data centers with some fascinating insights - Unit Economics: Compared to a traditional data center, AI factories have significant capex per MW need (5-10x higher) but still come out with a better IRR potential. - Asset Pivot: Unlike traditional data centre, AI factories means ownership of shell (land, infra, power) and semiconductors. GPU racks, Memory all are expensive and has a meaningful asset value. Runs a risk that lower utilization results in significant IRR hit. - Power demand: 15-20% of operating cost of AI factories is power. As MS expects AI factories to grow at a CAGR of 81% vs 10% in traditional data centers, it reflects in power demand. From 2027, MS expects AI factories to have almost same demand as old data centers. - Rent or own: Capex requirement is so high that there is a place for renting AI factories. Neoclouds enable that. They typically install GPU/TPUs after owning/leasing them. They finance this through SPVs from customers or investors. The revenue per MW though is lesser than that of hyperscalers. The key risks foreseen are: - Server depreciation before the breakeven point - Hyperscalers and frontier labs reneging on their demand All underlying numbers suggest demand isn't dying and business model doesn't look as hopeless as bear scenario suggests. Which other risk do you foresee?

glocalinvestor's tweet photo. Morgan Stanley issued a very interesting report on the AI data centers with some fascinating insights

- Unit Economics: Compared to a traditional data center, AI factories have significant capex per MW need (5-10x higher) but still come out with a better IRR potential.

- Asset Pivot: Unlike traditional data centre, AI factories means ownership of shell (land, infra, power) and semiconductors. GPU racks, Memory all are expensive and has a meaningful asset value. Runs a risk that lower utilization results in significant IRR hit.

- Power demand: 15-20% of operating cost of AI factories is power. As MS expects AI factories to grow at a CAGR of 81% vs 10% in traditional data centers, it reflects in power demand. From 2027, MS expects AI factories to have almost same demand as old data centers.

- Rent or own: Capex requirement is so high that there is a place for renting AI factories. Neoclouds enable that. They typically install GPU/TPUs after owning/leasing them. They finance this through SPVs from customers or investors. The revenue per MW though is lesser than that of hyperscalers.

The key risks foreseen are:
- Server depreciation before the breakeven point
- Hyperscalers and frontier labs reneging on their demand

All underlying numbers suggest demand isn't dying and business model doesn't look as hopeless as bear scenario suggests.

Which other risk do you foresee?

25

398

70

511

58K

0

2

0

72

Sean White

@SeanWhite_NTSB

6 days ago

🤦‍♂️ @EricKno67379868

Andrew Curran

@AndrewCurran_

6 days ago

Bernie Sanders and Greg Casar today announced the Ban Artificial Superintelligence Act. All AI development in the United States will be paused. Systems that have capabilities that match or exceed human cognitive performance will be banned. Violators will face 20 years in prison.

AndrewCurran_'s tweet photo. Bernie Sanders and Greg Casar today announced the Ban Artificial Superintelligence Act. All AI development in the United States will be paused. Systems that have capabilities that match or exceed human cognitive performance will be banned. Violators will face 20 years in prison. https://t.co/B6AhM3GiZl

1K

4K

403

1K

2M

0

1

0

17

Sean White

@SeanWhite_NTSB

6 days ago

The one on the left just looks more refined / more elegant. Also, if I'm a repair technician of such robot hands, the one of the left looks more appealing to me to actually work on...

Mustafa.

@oprydai

7 days ago

robotic hands are converging on two very different mechanical philosophies: tendon driven actuation and putting the actuators directly inside the fingers. tendon driven hands behave more like biology. motors can sit farther away from the joints and transmit force through cables, keeping the fingers relatively light while allowing compliant, organic motion. but now you inherit routing, tensioning, friction, stretch, backlash and maintenance problems. the hand becomes mechanically elegant but systemically complicated. in finger actuation attacks the problem from the opposite direction. put compact motors, gearboxes and sensing close to each joint and suddenly the transmission path becomes short and explicit. control becomes more direct, modularity improves and individual fingers can become self contained mechanisms. but you pay for that simplicity somewhere else: mass moves into the fingers, packaging becomes brutal, thermal management gets harder, impact loads reach the actuators more directly, and every millimeter of volume matters. this is why robotic hands are such a good representation of robotics as a whole. there is rarely a universally superior architecture. you are moving complexity around. tendons move complexity into transmission and control. distributed actuators move complexity into packaging, power density and thermal design. the winning humanoid hand probably won’t be the one with the most degrees of freedom. it will be the architecture that finds the best compromise between dexterity, force, compliance, weight, reliability, manufacturability and cost.

oprydai's tweet photo. robotic hands are converging on two very different mechanical philosophies: tendon driven actuation and putting the actuators directly inside the fingers. tendon driven hands behave more like biology. motors can sit farther away from the joints and transmit force through cables, keeping the fingers relatively light while allowing compliant, organic motion. but now you inherit routing, tensioning, friction, stretch, backlash and maintenance problems. the hand becomes mechanically elegant but systemically complicated.

in finger actuation attacks the problem from the opposite direction. put compact motors, gearboxes and sensing close to each joint and suddenly the transmission path becomes short and explicit. control becomes more direct, modularity improves and individual fingers can become self contained mechanisms. but you pay for that simplicity somewhere else: mass moves into the fingers, packaging becomes brutal, thermal management gets harder, impact loads reach the actuators more directly, and every millimeter of volume matters.

this is why robotic hands are such a good representation of robotics as a whole. there is rarely a universally superior architecture. you are moving complexity around. tendons move complexity into transmission and control. distributed actuators move complexity into packaging, power density and thermal design. the winning humanoid hand probably won’t be the one with the most degrees of freedom. it will be the architecture that finds the best compromise between dexterity, force, compliance, weight, reliability, manufacturability and cost.

25

596

74

302

27K

0

29

Sean White

@SeanWhite_NTSB

6 days ago

And do it fast

a16z @a16z

8 days ago

a16z's tweet photo. https://t.co/8DvRs1GJGq

234

26K

3K

4K

2M

0

22

SeanWhite_NTSB retweeted

Nic Cruz Patane

@niccruzpatane

8 days ago

The @Tesla Cybercab is launching this week. Good luck to Waymo.

530

13K

1K

982

1M

Sean White

@SeanWhite_NTSB

6 days ago

Amazing @EricKno67379868 @BlakeAMcKibbin @stbeden

Brett Adcock

@adcock_brett

6 days ago

Today, Figure is partnering with Nscale to deploy up to 100,000 GPUs on the NVIDIA Vera Rubin Platform We're committing $3.5 billion initially, with plans to scale beyond $6 billion To ship a robot into every home, we need a massive amount of compute

146

3K

245

346

2M

0

24

Sean White

@SeanWhite_NTSB

6 days ago

Elon Musk speaks to the G20 Summit - worth a watch: https://t.co/u11E8NFwHv @EricKno67379868 @BlakeAMcKibbin @stbeden

0

2

0

43

Sean White

@SeanWhite_NTSB

8 days ago

🤔

QuantumAnalysis

@Qtumanalysis

9 days ago

IN FOUR MONTHS, THE NSA STOPS BUYING ANYTHING THAT RUNS ON TODAY'S ENCRYPTION. THE REASON IS QUANTUM COMPUTERS. This is not a proposal, but a published directive with a date. It's called CNSA 2.0, the NSA issued it in 2022, tightened it in 2025, and almost nobody outside the agency has read it. It quietly orders every national security system in America to abandon the encryption the entire internet runs on, because a large enough quantum computer breaks all of it, and the NSA has decided that machine is coming. The schedule: January 1, 2027: every new system the government buys for classified or sensitive work must use quantum resistant encryption. 2030: all network equipment and software signing must run quantum resistant only. Legacy gear that can't be upgraded gets ripped out. 2033: operating systems, applications, cloud. Everything. 2035: full quantum resistance across all national security systems. Hard stop. The NSA gave the EARLIEST deadlines to two categories, signed software and network traffic. Their stated reason: those are most exposed to harvest-now-decrypt-later, adversaries recording encrypted data today to crack it once the quantum computer exists. Defense systems take 18 to 36 months to build. Anything designed today ships after the deadline. For much of the government, the planning window already closed. Meanwhile your bank, your hospital, and the app you're reading this on run the exact algorithms the NSA is abandoning. $INFQ $IONQ $QBTS $RGTI

Qtumanalysis's tweet photo. IN FOUR MONTHS, THE NSA STOPS BUYING ANYTHING THAT RUNS ON TODAY'S ENCRYPTION.

THE REASON IS QUANTUM COMPUTERS.

This is not a proposal, but a published directive with a date.

It's called CNSA 2.0, the NSA issued it in 2022, tightened it in 2025, and almost nobody outside the agency has read it. It quietly orders every national security system in America to abandon the encryption the entire internet runs on, because a large enough quantum computer breaks all of it, and the NSA has decided that machine is coming.

The schedule:

January 1, 2027: every new system the government buys for classified or sensitive work must use quantum resistant encryption.

2030: all network equipment and software signing must run quantum resistant only. Legacy gear that can't be upgraded gets ripped out.

2033: operating systems, applications, cloud. Everything.

2035: full quantum resistance across all national security systems. Hard stop.

The NSA gave the EARLIEST deadlines to two categories, signed software and network traffic. Their stated reason: those are most exposed to harvest-now-decrypt-later, adversaries recording encrypted data today to crack it once the quantum computer exists.

Defense systems take 18 to 36 months to build. Anything designed today ships after the deadline. For much of the government, the planning window already closed.

Meanwhile your bank, your hospital, and the app you're reading this on run the exact algorithms the NSA is abandoning.

$INFQ $IONQ $QBTS $RGTI

30

350

103

230

24K

0

18

Sean White

@SeanWhite_NTSB

11 days ago

Well said and hopeful @EricKno67379868 @GlennMcCJr @BlakeAMcKibbin

Brivael Le Pogam

@brivael

11 days ago

Aujourd’hui, le Far West est dans l’espace. Un type à la vie assez riche, assez développée, m’a dit un jour qu’on était trop sur Terre. Pas un militant. Pas un écolo de campus. Quelqu’un qui a déjà tout. Une vie pleine. Et cette phrase, lâchée comme une évidence. Ce n’est pas une observation. C’est un logiciel. Le même que celui des globalistes : on est trop nombreux. Il faut limiter le développement des humains et des animaux, sinon on va consommer toutes les ressources. C’est du Malthus. Pur. En 1798, Malthus annonce que la population va croître plus vite que la nourriture. Famines. Effondrement. Il s’est trompé totalement. Parce qu’il a traité l’humain comme une bouche. Pas comme un cerveau. La ressource ultime, ce n’est pas le blé. Ce n’est pas le pétrole. C’est l’être humain. Plus il y a d’esprits qui cherchent, plus on invente, plus la rareté recule. Depuis Malthus, on est passé de 800 millions à 8 milliards. On mange mieux. On vit plus longtemps. Les ressources sont plus abondantes, pas moins. Ehrlich a parié contre Simon sur l’épuisement. Il a perdu. Le Club de Rome a recyclé la même peur. Elle a encore perdu. Le logiciel, lui, n’a pas bougé. Ce mindset, dans la tête de beaucoup de bourgeois et de beaucoup d’élites, est extinctionniste. Et profondément nihiliste. Il nous amène à notre perte. Elon a posé le vrai cadre. Aucune civilisation n’est parfaitement stable. Soit elle s’étend, soit elle se contracte. Un organisme qui arrête de croître s’éteint. Les gens ne comprennent pas que décroître la population humaine, ce n’est même pas un projet qui servira les élites. Si on décroît, on s’anéantit. Les retraites, l’innovation, la défense, la culture — tout ça tient sur le nombre de ceux qui construisent demain. Un monde plus petit n’est pas un monde plus contrôlable. C’est un monde qui meurt. C’est pour ça qu’il est vital de comprendre le projet d’Elon. Pas parce qu’il est riche. Parce qu’il est celui qui refuse le script de l’extinction. Multiplanétaire. Plus d’humains. Plus de conscience. Plus d’horizons. C’est très raccord avec ce qu’on a toujours fait. Trouver de nouveaux territoires. Hier le Far West. Aujourd’hui l’espace. Les extinctionnistes se trompent toujours pour la même raison. Ils figent le monde. Ils pensent que le gâteau est fixe. Ils oublient que l’humain agrandit le gâteau. Et ce projet-là est extrêmement dangereux parce qu’il se vend comme de la sagesse. Comme de la responsabilité. Comme de l’amour de la planète. Alors qu’il est un suicide civilisé. Un suicide qui commence par ceux qui y croient. Natalité en chute. Jeunesse sans futur désirable. Continent qui a cessé de faire des enfants parce qu’il a cessé de croire en demain. Les mauvaises idées tuent d’abord ceux qui y croient. La reconversion est possible. Le 21e siècle n’appartient pas à ceux qui gèrent la pénurie. Il appartient à ceux qui ouvrent le prochain Far West. Allez construire.

20

282

63

35

25K

0

1

0

29

SeanWhite_NTSB retweeted

Brivael Le Pogam

@brivael

11 days ago

Aujourd’hui, le Far West est dans l’espace. Un type à la vie assez riche, assez développée, m’a dit un jour qu’on était trop sur Terre. Pas un militant. Pas un écolo de campus. Quelqu’un qui a déjà tout. Une vie pleine. Et cette phrase, lâchée comme une évidence. Ce n’est pas une observation. C’est un logiciel. Le même que celui des globalistes : on est trop nombreux. Il faut limiter le développement des humains et des animaux, sinon on va consommer toutes les ressources. C’est du Malthus. Pur. En 1798, Malthus annonce que la population va croître plus vite que la nourriture. Famines. Effondrement. Il s’est trompé totalement. Parce qu’il a traité l’humain comme une bouche. Pas comme un cerveau. La ressource ultime, ce n’est pas le blé. Ce n’est pas le pétrole. C’est l’être humain. Plus il y a d’esprits qui cherchent, plus on invente, plus la rareté recule. Depuis Malthus, on est passé de 800 millions à 8 milliards. On mange mieux. On vit plus longtemps. Les ressources sont plus abondantes, pas moins. Ehrlich a parié contre Simon sur l’épuisement. Il a perdu. Le Club de Rome a recyclé la même peur. Elle a encore perdu. Le logiciel, lui, n’a pas bougé. Ce mindset, dans la tête de beaucoup de bourgeois et de beaucoup d’élites, est extinctionniste. Et profondément nihiliste. Il nous amène à notre perte. Elon a posé le vrai cadre. Aucune civilisation n’est parfaitement stable. Soit elle s’étend, soit elle se contracte. Un organisme qui arrête de croître s’éteint. Les gens ne comprennent pas que décroître la population humaine, ce n’est même pas un projet qui servira les élites. Si on décroît, on s’anéantit. Les retraites, l’innovation, la défense, la culture — tout ça tient sur le nombre de ceux qui construisent demain. Un monde plus petit n’est pas un monde plus contrôlable. C’est un monde qui meurt. C’est pour ça qu’il est vital de comprendre le projet d’Elon. Pas parce qu’il est riche. Parce qu’il est celui qui refuse le script de l’extinction. Multiplanétaire. Plus d’humains. Plus de conscience. Plus d’horizons. C’est très raccord avec ce qu’on a toujours fait. Trouver de nouveaux territoires. Hier le Far West. Aujourd’hui l’espace. Les extinctionnistes se trompent toujours pour la même raison. Ils figent le monde. Ils pensent que le gâteau est fixe. Ils oublient que l’humain agrandit le gâteau. Et ce projet-là est extrêmement dangereux parce qu’il se vend comme de la sagesse. Comme de la responsabilité. Comme de l’amour de la planète. Alors qu’il est un suicide civilisé. Un suicide qui commence par ceux qui y croient. Natalité en chute. Jeunesse sans futur désirable. Continent qui a cessé de faire des enfants parce qu’il a cessé de croire en demain. Les mauvaises idées tuent d’abord ceux qui y croient. La reconversion est possible. Le 21e siècle n’appartient pas à ceux qui gèrent la pénurie. Il appartient à ceux qui ouvrent le prochain Far West. Allez construire.

20

282

63

35

25K

SeanWhite_NTSB retweeted

Saganism

@Saganismm

12 days ago

“The ideal subject of totalitarianism... is people for whom the distinction between fact and fiction, true and false, no longer exists.” — Hannah Arendt

Saganismm's tweet photo. “The ideal subject of totalitarianism... is people for whom the distinction between fact and fiction, true and false, no longer exists.”

— Hannah Arendt https://t.co/6mUsKnWwbW

63

4K

2K

426

113K

Sean White

@SeanWhite_NTSB

11 days ago

True @EricKno67379868 @BlakeAMcKibbin

Jordan Bramble

@jordanbramble

12 days ago

This nails the @AntaresNuclear thesis spot on. The next gen nuclear that ultimately comes down the cost curve and reaches commercially performant economics will start with military deployment first.

0

39

6

3

2K

0

1

0

67