Anthony Pompliano

Should You Invest In SpaceX IPO, Elon Musk, Bitcoin or AI?

🇬🇧 EN🇪🇸 ES
54:48 min youtube 2026 Semana 24 🇪🇸 ES

Resumen

TL;DR

  • La idea central de Visser es que SpaceX no es solo una empresa de cohetes: combina negocios actuales con flujo real, como Starlink, con una opción gigantesca sobre la próxima ola de infraestructura de IA, sobre todo si Elon sigue comprimiendo tiempos de construcción de centros de datos y termina empujando parte del cómputo a órbita.
  • La lectura más útil para invertir no es “compra SpaceX a cualquier precio”, sino sigue los cuellos de botella físicos: Visser insiste en plata, cobre, energía, baterías, fosfuro de indio, químicos críticos y capacidad industrial doméstica como insumos escasos que la narrativa de IA todavía no puede saltarse.
  • Sobre los modelos de IA es bastante directo: el precio por token tenderá a deflación mientras los costes de cómputo y las restricciones de capacidad siguen siendo brutales. Para él eso abre un mid-cycle slowdown de 3–6 meses en los nombres más abarrotados de infraestructura IA y un ajuste más largo de 3–4 años en el que no todas las compañías de frontier models van a sobrevivir como ganadoras independientes.

◆ SpaceX se está vendiendo como dos compañías al mismo tiempo

El arranque de Visser es bastante concreto. Dice que Elon ya tiene un negocio real más cercano a un hyperscaler de lo que muchos inversores reconocen: Colossus de xAI supuestamente se levantó en 122 días, Anthropic y Google ya aparecen como clientes o socios dentro del stack que describe, y Starlink ya es un producto útil hoy para conectividad resiliente. Encima de eso coloca la parte especulativa: si SpaceX consigue seguir reduciendo fricción en lanzamientos, densidad satelital y despliegue de cómputo, la compañía tendría una opción enorme sobre el futuro de infraestructura IA que los modelos tradicionales todavía no saben valorar bien.

â–¶ El argumento del data center orbital es, en realidad, un argumento sobre cuellos de botella

Los números más claros de la charla aparecen aquí. Visser dice que un data center terrestre de 1 gigavatio cuesta alrededor de $60B, con unos $35B en chips y aproximadamente $25B en el resto de la construcción. A partir de ahí, usa el framing orbital de Elon para sugerir que esa parte no-chip podría caer hacia $5B si se eliminan grandes fricciones de tierra, refrigeración y energía. No lo presenta como algo ya demostrado. Lo presenta como el mismo patrón de Elon de siempre: detectar el cuello de botella físico, restar pasos y adueñarse de un camino hardware que nadie más puede replicar con facilidad.

★ Por qué aun así no termina en un “compra la IPO sin pensar”

Visser separa claramente la admiración por la narrativa de la construcción de cartera. Dice que SpaceX todavía necesita ejecutar muchas cosas de baja probabilidad para que la valoración actual se vea obvia desde aquí. Por eso su preferencia personal gira más hacia los insumos necesarios si toda la visión avanza de todos modos. Si los data centers orbitales, los humanoides y el cómputo agentic tardan más de lo que esperan los toros, los dueños de commodities e infraestructura aún podrían ganar mientras el sueño se reprecifica a un ritmo más lento.

◆ La trade física: plata, cobre, energía, baterías

Aquí la conversación se vuelve más operativa. Visser dice que quiere exposición a plata porque tanto las baterías como la infraestructura de data centers orbitales tiran de ella más de lo que muchos inversores entienden. También repite una estadística de materiales según la cual el mundo podría necesitar una cantidad de cobre en los próximos 10–15 años comparable a todo lo extraído en los últimos 10.000 años. Más allá de si esa proporción exacta termina siendo perfecta, su mensaje es claro: la IA no es solo software; es minería, red eléctrica, química, almacenamiento y manufactura, y esos sistemas no escalan con slides ni con storytelling.

▶ China, minerales críticos y por qué el buildout puede chocar con un mid-cycle slowdown

Su advertencia más específica en supply chain gira alrededor del fosfuro de indio, que llama un insumo crítico para infraestructura óptica y de data centers vinculada a nombres como Marvell, Coherent y Lumentum. Sostiene que China controla la oferta, igual que controla otros flujos minerales críticos, y argumenta que el pulso geopolítico entre EE.UU. y China hace poco realista una curva de capex de IA limpia y continua. De ahí sale una de sus tesis tácticas principales: podríamos estar entrando en un AI mid-cycle slowdown, es decir, en una fase de 3–6 meses donde habrá tanta decepción como sorpresa positiva en semis, cooling e infraestructura abarrotada.

◆ La soberanía industrial no va solo de tierras raras

El desvío de Athanor suena raro al principio, pero encaja con la tesis. Pompliano y Visser usan una historia sobre creatina y químicos nitrogenados para señalar algo más amplio: el problema de dependencia de EE.UU. atraviesa química, insumos especializados y eslabones oscuros de la cadena de suministro, no solo los minerales que salen en titulares. Traducido: si la próxima década gira en torno a reconstruir capacidad industrial, los ganadores pueden estar bastante más arriba en la cadena que las marcas obvias de IA.

★ Las compañías de modelos enfrentan una combinación incómoda: presión en precios y presión en capacidad

En la parte software, Visser cree que el mercado sigue subestimando lo feas que pueden ponerse las unit economics antes de mejorar. Cita dos anclas: análisis tipo SemiAnalysis que sugieren que las suscripciones siguen subsidiadas frente al coste real del token, y la idea de que 80–85% de las queries podrían moverse fuera de los modelos frontier más caros. Al mismo tiempo, dice que el coste de producir tokens está alrededor de 4x por encima del año pasado mientras la capacidad sigue restringida. Por eso piensa que los grandes labs están corriendo a mercado, bajando precios y buscando mejores tiers de monetización al mismo tiempo.

▶ Su visión real de la guerra de modelos: viene comoditización

La postura de Visser no es “la IA es una burbuja”. Es más fina: la adopción sigue muy temprano, la demanda sigue explotando, pero los modelos en sí probablemente se van a comoditizar. Ve en DeepSeek, en alternativas open source más baratas y en el routing hacia modelos locales o más baratos una presión estructural sobre el pricing premium. Espera una industria más parecida a las aerolíneas: inferencia cara para casos de alto valor, modelos baratos para tareas normales y mucha más disciplina en eficiencia de tokens dentro de las empresas.

◆ El papel de crypto es verificación, escasez y comunidad en un mundo lleno de slop

La parte de los Knicks suena personal, pero vuelve directo a su tesis de crypto. Visser argumenta que en un mundo lleno de AI slop, deepfakes y contenido sintético, las blockchains importan porque pueden probar asistencia real, procedencia real y escasez real. Su ejemplo es simple: un ticket conmemorativo físico puede falsificarse en cuanto tenga valor; una prueba on-chain de asistencia y propiedad viaja con el fan para siempre. Por eso sigue agrupando stablecoins, tokenización y NFTs como utilidad real de crypto y no como nostalgia especulativa.

◆ Bezos, physical AI y hacia dónde se mueve el talento top

El segmento rápido sobre Bezos refuerza el mismo régimen. Visser dice que si la IA permite hacer 80–90% del trabajo con 10–20% de la gente, la siguiente frontera obvia ya no son más demos abstractas de software, sino manufactura física, robótica, farma y cuellos de botella del mundo real. Interpreta el nuevo empuje industrial de Bezos como otra señal de que los mejores operadores quieren controlar la capa hardware y de producción, porque ahí sigue habiendo escasez, inflación y margen.

â—† Buscar el alpha

La lectura más limpia es que Visser no está diciendo que abandones IA, Bitcoin o SpaceX. Está diciendo que se termina la fase perezosa del trade. La siguiente pierna pertenece a quien sepa distinguir entre equity de sueño y insumos escasos, entre modelos premium e inferencia comoditizada, y entre narrativas digitales y cuellos de botella físicos.

  • Caso alcista de SpaceX: motores de ingresos actuales más una gran opción sobre lanzamientos, cómputo e infraestructura orbital si Elon sigue rompiendo cuellos de botella físicos.
  • Expresión preferida: Visser suena más cómodo con plata, cobre, energía, baterías y materiales críticos que pagando cualquier valoración por el sueño directamente.
  • Riesgo IA a corto plazo: un mid-cycle slowdown de 3–6 meses en infraestructura IA abarrotada es plausible porque las expectativas corrieron mucho más rápido que la realidad de la supply chain.
  • Riesgo para las model companies: tokens subsidiados, costes de cómputo al alza, límites internos al consumo y alternativas open source más baratas apuntan a compresión de márgenes y comoditización.
  • Utilidad de crypto: aquí el caso blockchain es prueba, procedencia y escasez digital, sobre todo cuando la IA abarata fabricar contenido falso.
  • Apuesta de régimen a largo plazo: los ganadores duraderos pueden ser las empresas que ayuden a resolver energía, química, materiales, logística, almacenamiento y manufactura.
Tema Ancla de la charla Por qué importa
SpaceX / cómputo orbital 122 días para Colossus; SpaceX descrita como nuevo hyperscaler; parte no-chip planteada como $25B en tierra vs. $5B en órbita El caso alcista depende de que Elon elimine cuellos de botella físicos más rápido que los incumbentes
Insumos críticos Plata, cobre, energía, baterías aparecen una y otra vez como restricciones duras Pueden ganar incluso si las grandes narrativas de equity IA se frenan o se reratean
Geopolítica de la supply chain Fosfuro de indio y otros minerales clave bajo control chino La ruta del capex IA no es geopolíticamente limpia
Economía de modelos 80–85% de queries quizá no necesiten modelos top; coste de producir tokens cerca de 4x YoY El pricing premium sufre presión por demanda y por oferta a la vez
Crypto / NFTs El ejemplo del ticket de los Knicks se usa para defender prueba on-chain de asistencia y autenticidad La IA aumenta el valor de la procedencia verificable
Physical AI Bezos / manufactura / humanoides como siguiente dirección principal El cuello de botella se está moviendo del código al mundo real
El giro: La inversión más importante de toda la conversación es que Visser trata la IA primero como una historia de materiales e infraestructura, y solo después como una historia de software. Si ese framing es correcto, el mejor risk/reward puede estar menos en los activos narrativos más ruidosos y más en los cuellos de botella aburridos que todos esos activos todavía necesitan para funcionar.

► Resumen por capítulos

1. Intro (0:00)

Pompliano prepara una conversación sobre SpaceX, infraestructura IA, data centers orbitales, crypto y physical AI. El hilo conductor es cómo posicionar una cartera cuando la demanda de cómputo sigue chocando con restricciones del mundo real.

2. SpaceX IPO, Elon & orbital data centers (1:00)

Visser plantea SpaceX como flujo presente más una opción enorme sobre futura infraestructura de cómputo. Su tesis es que la ventaja de Elon no es solo visión, sino capacidad repetida para eliminar cuellos de botella de energía, cooling, launches y velocidad de ejecución.

3. Critical minerals & the AI supply chain (12:15)

La conversación pasa del activo glamuroso a los insumos. Plata, cobre, energía y almacenamiento aparecen como piezas donde el mercado puede seguir infraponderado si solo persigue los nombres obvios de IA.

4. American industrial sovereignty & critical chemicals (18:55)

Una anécdota sobre creatina y químicos nitrogenados termina en una tesis de reindustrialización. El punto es que la dependencia estratégica va mucho más allá de las tierras raras que ya salen en titulares.

5. AI model cost crisis & who wins the model war (22:36)

Esta es la sección más directamente de mercado. Visser dice que la adopción puede seguir explotando mientras empeoran las economics de las model companies, porque el cómputo sigue restringido y la inferencia premium puede comoditizarse antes de lo que espera el consenso.

6. Knicks NBA Finals & the case for crypto in an AI world (37:10)

El desvío deportivo acaba siendo un argumento pro-crypto: en un mundo de falsificaciones y contenido sintético, la prueba de asistencia, la escasez y la procedencia verificable ganan valor.

7. Jeff Bezos launches Prometheus (47:48)

El empuje industrial de Bezos se interpreta como otra señal de que la siguiente frontera es la IA física. Si la capa software sigue mejorando, el valor se desplaza hacia ejecución real y eliminación de restricciones físicas.

8. AI use case of the week (50:34)

El cierre muestra construcción de producto asistida por IA en tiempo real durante una llamada con cliente. Visser lo usa para reforzar dos ideas: que el comportamiento AI-native importa más que los silos tradicionales y que la desaceleración actual convive con un bull market más largo en uso real de IA.

Generado con algoritmo v2.1-anchor-first · modelo openai-codex/gpt-5.4 · 2026-06-14T11:06:11Z

Transcripción

[0:00] And that has always been my belief is
[0:01] that these models will be commoditized,
[0:03] that you don't need to use the most
[0:05] expensive ones, and that the
[0:07] corporations haven't figured out yet how
[0:08] to use them. When you put all that
[0:10] together and then you add in Bernie
[0:12] Sanders and Donald Trump want to own
[0:13] part of the companies, I I posted on X,
[0:16] there are so many headwinds when the
[0:17] Pope speaks out against the models.
[0:19] Like, how many things do you want to
[0:20] have where people are like, "Slow this
[0:22] down." And then, what's going on, guys?
[0:24] Today, we got a great conversation with
[0:26] Jordy Visser. In this conversation, we
[0:27] talk about the SpaceX IPO, what he likes
[0:29] about it, where he thinks there could be
[0:30] risk, and what you should think about in
[0:32] your portfolio. On top of that, we talk
[0:34] about artificial intelligence, orbital
[0:35] data centers, what's going on with the
[0:36] model companies, token cost, and usage
[0:39] of these products. And then we get into
[0:40] physical AI and robotics, what Jeff
[0:42] Bezos is doing, how Jord is thinking
[0:44] about all these different tools. And
[0:45] last but not least, we talk about the
[0:46] New York Knicks, the future NBA
[0:49] champions, but we don't talk about it
[0:51] from a basketball perspective. We talk
[0:52] about it as how it relates to AI in
[0:54] crypto. and I think it'll give you a lot
[0:56] to think about as to where the future is
[0:58] going. Here's my latest conversation
[0:59] with Jordi Visser. All right, Jordy,
[1:01] let's talk about the SpaceX IPO.
[1:03] Yesterday, it goes public. Everyone's
[1:04] super excited about it. What's your
[1:06] general take on the company valuation,
[1:08] whether investors should look at this as
[1:10] a space company, an AI company? Just how
[1:12] do you evaluate Elon's great
[1:14] accomplishment here?
[1:16] >> Um, so I'll break it into two
[1:18] components, the now and the future. Um,
[1:21] so think of it as the same way I think
[1:23] of AI and Bitcoin, meaning AI and
[1:26] crypto. We're in the agentic side.
[1:28] Everything matters about that. So what
[1:29] do they have in the agentic side that
[1:31] matters? Well, they've obviously got the
[1:33] Colossus side. I think they're now the
[1:34] fourth largest hyperscaler. So I just
[1:37] want to like say to everyone listening
[1:39] >> no 30 days 45 days meaning they didn't
[1:44] have any deals and now they have
[1:45] anthropic and Google
[1:47] >> and his ability to do the first colossus
[1:51] was 122 days on something that is
[1:54] normally taking people years to get
[1:56] done. So Elon is the master of
[1:59] engineering and the ability to um
[2:02] vertically integrate all of this stuff
[2:03] but he also figures out solutions along
[2:05] the way. So, if anyone's going to be
[2:06] able to build data centers faster than
[2:08] all of the chaos we're seeing, I go back
[2:10] to something I said here, which Elon
[2:12] said publicly, which was when you get
[2:14] software guys trying to do hardware.
[2:16] It's a very different game. So, he's a
[2:17] hardware guy who's involved in the
[2:20] software game from the AI side with XAI.
[2:23] Remember, he I'm going to leave the the
[2:25] AI side alone for a second. Starlink is
[2:27] the other part that is today. So,
[2:30] Starlink is really important. I have a
[2:32] Starlink up in Maine. And the reason I
[2:33] have it in Maine is because I'm in a
[2:35] very quiet place where there is cable,
[2:37] but it goes out all of the time. And so
[2:40] by having Starlink as a backup choice,
[2:42] it allows me to do this with you during
[2:43] the summer in Maine if there's a storm.
[2:45] And I have a feeling with all the
[2:46] humidity that's already in the arc,
[2:47] there's going to be storms this year
[2:48] that weren't last year. Um, so Starlink
[2:50] gives me the capability of doing this
[2:51] any time. But the ability to expand that
[2:54] and the revenues, those are today's
[2:55] businesses with the Colossus thing being
[2:57] massive in terms of the dollars. The
[3:00] future for SpaceX is space stations. The
[3:03] future is all of these different parts
[3:06] of being able to deal with the rocket
[3:09] ships coming back and then going out the
[3:10] same day, which if he gets it done next
[3:13] year, great. This kind of is like the
[3:15] Tesla part. So, if you think about his
[3:17] business, he now has all of a sudden out
[3:19] of nowhere this component part which is
[3:22] revenues coming in the door in a
[3:24] business where it makes him the fourth
[3:26] largest hyperscaler. I mean, it's really
[3:27] hard to describe that he's joined in
[3:29] with Amazon, Microsoft, and Google
[3:31] already and he's getting business from
[3:33] one of them and he's getting business
[3:35] from Anthropic on that with Starlink
[3:37] >> and he did the deal with Cursor
[3:39] >> and that's the the next thing I was
[3:40] going to say is you get the AI part two.
[3:42] So, he's behind with XAI for sure, but
[3:45] what Curser allows him to do is
[3:47] accelerate the coding side. And so, that
[3:50] has already helped. And remember, he
[3:53] gets to plug into the same thing that
[3:54] Anthropic and Google are plugging into,
[3:57] which again was the first real thing of
[4:00] Blackwells. Now, from what I've heard,
[4:02] and I heard this on a podcast recently,
[4:04] he has secured 20% of all of the Vera
[4:06] Rubin that'll be going out. Now,
[4:08] >> interesting.
[4:08] >> The reason that's important is because
[4:10] if Jensen was going to give these to
[4:13] anyone, these racks, he'd go with the
[4:15] person that's already proven that he can
[4:17] build the data center in this type of
[4:18] environment. And that's the part that I
[4:20] don't think people have have have fully
[4:22] figured out is I think the premium that
[4:24] should come for SpaceX is the ability of
[4:27] him to navigate the bottlenecks that
[4:29] we're seeing and be able to build these
[4:30] data centers both terrestrial where he's
[4:32] getting the side. But he also put out
[4:34] the information on what the space
[4:36] stations will be. So when you break it
[4:38] down, uh a data center right now for 1
[4:40] gawatt is like 60 billion of of of that
[4:42] about 35 billion is the chips. So you're
[4:45] left with 25 billion. and he's saying he
[4:47] can do the um space station
[4:51] for 5 billion.
[4:54] 5 billion versus 25 billion. The chips
[4:56] are still going to be the same. That's a
[4:57] dramatic difference, which means the
[4:59] margins on his compute would be up
[5:00] there. So, if he's able to do that, and
[5:03] I think Google's invested in that, too.
[5:05] Meaning the reason they did the deal is
[5:07] if he's able to do this, they want to
[5:09] have already secured the rights to that.
[5:11] So, who knows what's in the documents.
[5:12] So, I think SpaceX is really two
[5:14] companies. you're getting a put in it,
[5:17] which is the revenue that's coming in
[5:18] right now. It's still valued extremely
[5:20] rich, but then you're getting this
[5:21] massive call option on these other
[5:23] parts.
[5:23] >> So, the way I break down I think the
[5:25] investor conversations I'm seeing or
[5:27] debates, there's people who use
[5:28] spreadsheets and they're looking at you
[5:30] know SpaceX is 25 times more uh you know
[5:33] overvalued than they'll name some
[5:35] company. They'll look at uh first AI IPO
[5:37] performance versus the first 6 months or
[5:39] 12 months. uh every historical norm,
[5:42] every historical trend, every single
[5:44] data point in their spreadsheet is
[5:45] telling them SpaceX is overvalued. The
[5:47] other side of the debate are people who
[5:49] essentially believe in Elon and they're
[5:51] saying, "This guy has a track record of
[5:52] being successful. He's the best
[5:53] entrepreneur of our lifetime. I'm
[5:55] willing to bet that just like with
[5:56] Tesla, which is up 25,000%, has
[5:59] compounded annually at 40% a year since
[6:01] he went public. He's going to figure it
[6:02] out and he's going to be able to drive
[6:03] returns for investors." So, if that's
[6:05] the debate, which I I generally think it
[6:06] is, if you then go look at like orbital
[6:08] data centers, I hear it in a very micro
[6:10] way where people say, "Oh, I'm going to
[6:12] take these same exact chips. I'm going
[6:13] to put them into space." A bunch of
[6:14] people say, "Orbital data centers plus
[6:16] Elon Musk, I'm in." And then I hear
[6:18] people say, "What happens if the chips
[6:21] get uh need to be uh refurbished or
[6:24] there's an issue? Uh, we don't have
[6:26] humans there, so I don't I'm not going
[6:28] to invest because, you know, we haven't
[6:30] solved that problem yet." And again, it
[6:32] goes back to I think that there's this
[6:33] like rational spreadsheet driven kind of
[6:37] approach and then there is just like I'm
[6:38] betting on the person.
[6:40] >> How do you balance those two things?
[6:42] >> Well, there's a lot in there. Uh so I
[6:45] think the problem and and I've started
[6:47] to talk more and more about this.
[6:49] Building data centers on land might be
[6:52] at this point more complicated than
[6:54] doing them in space. And it's very hard
[6:56] for people to go through this. So
[6:57] >> because
[6:58] >> because of the amount of complexity
[7:00] there is with all of the components that
[7:03] the cooling which is not needed in
[7:04] space. Like when you go through all of
[7:06] the components that go from 25 billion
[7:08] to 5 billion, you're reducing a lot of
[7:10] the friction. But the other thing you're
[7:11] reducing where do we get the land from?
[7:13] Where do we get the power from? This is
[7:15] why he wants to do it in space. Let's
[7:17] just take two parts. The cooling and the
[7:20] energy. Okay? The sun's producing the
[7:22] energy and the cooling we don't need. If
[7:25] you go through that and then you add in
[7:26] the government regulations, the push
[7:28] backs that are coming from we can't do
[7:29] this in the land, I find it very hard to
[7:32] believe that that path is going to be
[7:33] anywhere near as smooth as all of these
[7:35] charts that the sell side puts out on a
[7:37] regular basis. Now, I'm a person that
[7:39] believes in the token usage and believes
[7:41] that we will get to the gigawatts
[7:42] necessary. If you ask me, the two ways
[7:44] we're going to get there, one is going
[7:45] to be batteries. We have to have more
[7:47] storage because I don't believe the
[7:50] government's going to allow us and the
[7:52] the lack of plumbers, electricians, the
[7:54] bottlenecks that are happening at all
[7:56] different components. Memory is going to
[7:57] allow us to get done faster. I think the
[7:59] reality is if we can store energy off
[8:01] the grid, then we can get to the point
[8:04] where we already have enough on the grid
[8:05] on the land. It's going to take a while
[8:07] for the the batteries. All of this is a
[8:09] game of how quick can the innovation
[8:10] happen to get us there. So, when you're
[8:12] betting Elon, you are betting on the
[8:13] person. You're betting on this. But I
[8:16] just want to make one more thing clear
[8:17] for people. So all of last year when I
[8:20] was talking about Micron,
[8:22] it was a $60 to $110 stock between
[8:25] January and let's say April May even
[8:29] even in May.
[8:31] The reason people didn't want to own it
[8:33] is because they didn't think the
[8:35] earnings would happen. Okay. So then the
[8:38] stock goes to 700, 800, 900, a,000. How
[8:43] is something go up a,000%
[8:45] from where it was during the first
[8:47] quarter of last year and it's still
[8:50] cheap today? Well, for that to happen,
[8:53] everyone's forecast on the exponential
[8:55] needs were wrong. This is the issue is
[8:57] if SpaceX is able to send rocket ships
[8:59] up and back down and by the end of 28
[9:02] they're doing what he wants, which is
[9:05] thousands a year, which means at least
[9:07] three a day where the ship is going up
[9:09] and coming back. Then all of a sudden
[9:10] the amount of satellites that he has out
[9:12] there. Who else can do this? Who's the
[9:13] competition? That's part of what goes
[9:15] into a price. Can Google do that? They
[9:17] don't have the rocket ships to do that.
[9:18] Where are they going to secure the
[9:19] rocket ships? He has most of the
[9:20] satellites in space. So at some point
[9:22] you have to get into the challenges of
[9:24] doing things on land. Who the person is
[9:26] and has he proved that he can do things
[9:28] that are unthinkable and has he figured
[9:30] out a way to go through this? And we're
[9:31] all leaving out the fact that within the
[9:33] next 5 years he'll be using humanoids to
[9:35] do a lot of this stuff. So I just think
[9:37] it's a very hard thing for people to get
[9:38] in. But if the earnings come through
[9:40] even slightly above expectations and if
[9:42] he's sending rocket ships into space and
[9:44] they're coming back and then going back
[9:45] out in the same day, you're going to
[9:47] have a valuation change. Now, that's
[9:48] still on the distribution of outcome.
[9:50] So, I can see where people are bearish
[9:51] the name. I can see where they go
[9:52] through it. But Elon has proven that if
[9:54] you if you get uh if you go against him
[9:57] on the possibility, he has enough of a
[10:00] fan base and enough people that have
[10:01] believed him and made a lot of money off
[10:02] him that I think that bid's going to be
[10:04] underneath no matter what. The whole
[10:06] thing about um space and the ability to
[10:08] do this to me is uh reinforced. There's
[10:11] a book that I read. I know you don't
[10:12] read books but uh it's called the
[10:14] algorithm and it was written by the guy
[10:15] who became the I think the president of
[10:17] Tesla and he talks about the way he got
[10:19] hired was uh he got introduced by Cheryl
[10:21] Samberg. Elon basically is like I got a
[10:23] problem in our sales uh process. And so
[10:26] this guy goes to a bunch of Tesla
[10:28] dealerships, uses different emails and
[10:29] does some test drives. And he's trying
[10:31] to go through the sales process and
[10:32] figure out where are the bottlenecks.
[10:33] and he came from the manufacturing world
[10:35] and what he realized was the way you
[10:36] determine where the bottlenecks are is
[10:38] you look where the physical products
[10:40] start to pile up naturally go over there
[10:43] and figure out why is this piling up and
[10:45] unblock the the bottleneck.
[10:47] >> So he tells a story that at Tesla what
[10:49] happened is everyone was uh being
[10:50] compensated for test drives not for
[10:53] conversions. And so what happens you'd
[10:54] go and you do a test drive and no one
[10:56] would ever call you. So he calls up the
[10:58] guy who's head of sales and says, "Hey,
[11:00] no more test drives for a week. Call
[11:02] every single person back." All of a
[11:03] sudden sales explode, right? And they go
[11:05] and so he goes and he works there and
[11:07] Elon's way of hiring is, "Hey, you solve
[11:09] the problem." So like you must be able
[11:10] to solve other problems here. He talks
[11:12] about the algorithm which now is famous
[11:14] in Elon world. And the entire idea of
[11:16] the algorithm is you go and you find the
[11:17] bottleneck and then you undo the
[11:19] bottleneck and then you go to the next
[11:20] bottleneck and you undo it. But what he
[11:23] believes is that you should be
[11:24] subtracting as much as possible. when
[11:26] he's got this like idiot index of what
[11:28] are the cost of the inputs and what is
[11:29] the end product. What you just described
[11:32] is like the perfect implementation of
[11:34] the algorithm. One of the bottlenecks is
[11:37] energy. Well, what if we just get energy
[11:39] all the time from the sun?
[11:40] >> Mhm.
[11:40] >> One of the bottlenecks is all the
[11:41] cooling. What if we just remove that?
[11:43] Right? And he just keeps going through
[11:44] this. But I also think it's the other
[11:47] side of this which is he understands and
[11:49] is way more strategic than people I
[11:51] think give him credit for that he has a
[11:53] monopoly on rocket launches. Mhm.
[11:55] >> And so not only can he make his business
[11:57] successful, can he drive the cost down
[11:59] and and you know be competitive, but
[12:01] he's also going to do something that no
[12:02] one else can do. And so it provides this
[12:05] huge moat that only really exists in the
[12:09] hardware world going forward. Like the
[12:11] AI world, it's going to be very hard to
[12:12] build these modes, which you've talked
[12:13] about a bunch as well, right?
[12:15] >> Yeah. Here here's where I'll get um
[12:20] more negative on investing in SpaceX
[12:23] just on everything you described. So
[12:27] we love narratives of course it's what
[12:30] makes human beings happy sitting around
[12:32] the fire when we started eating meat and
[12:34] our brains started you know changing and
[12:36] we're telling stories and at the end of
[12:38] the day I'm a storyteller you're a
[12:39] storyteller like we tell stories but
[12:41] >> I invest
[12:42] >> got a fireplace here. Yeah, exactly. Um,
[12:45] the reason I chose Micron over other
[12:48] things last year and I remember talking
[12:51] to Phil Rosen about this when he put
[12:53] this in his his paper which was um so
[12:56] why Micron and I gave all the reasons
[12:58] for the next well to build the data
[13:00] centers and to move into the agentic
[13:02] world and the agentic world is necessary
[13:04] for humanoids down the road. This is a
[13:06] never- ending thing. We need memory. And
[13:08] I don't think people have really
[13:09] realized that eventually when we make
[13:10] that change into the vision of the
[13:12] agentic world, the amount of memory
[13:14] we're going to need is off the charts.
[13:16] So Micron is a component of that. Here's
[13:18] what I just said. Batteries on one side
[13:21] and
[13:22] orbital space stations for data centers.
[13:26] Do you know what both of them need a lot
[13:28] more of than the traditional one?
[13:30] Silver. I want more silver. So, I
[13:34] listened to a podcast this week from
[13:35] All-In at their liquidity event or
[13:37] whatever they're calling it, and they
[13:38] interviewed someone on the material
[13:40] side, and I highly recommend it's a
[13:41] 25minute interview. The reason it's 25
[13:44] minutes is I'm sure people won't care
[13:46] that much about it. I mean, if it was
[13:48] Bill Gurley or someone, he'd be up there
[13:50] for over an hour. Um what the guy said,
[13:53] okay, the amount of copper we need for
[13:55] the next 10 years is equivalent, and I
[13:57] think he said 10 years, but it's 10 or
[13:59] 15, it doesn't really matter for the
[14:00] point, is equivalent to all the copper
[14:02] mined over the last 10,000 years. So
[14:05] when people look at this problem, I'm
[14:07] looking for the parts that they're going
[14:08] to need to actually accomplish this cuz
[14:10] it might take him 20 years to get all
[14:13] Elon's been known to say it's going to
[14:15] happen sooner. It helps him raise more
[14:16] money. It helps him do things. So, let's
[14:18] assume it takes double the time for
[14:20] humanoids, for everything. Well, during
[14:21] that time, silver is going to be a
[14:22] bottleneck the entire time, and I think
[14:24] it'll go 10 times where it is today.
[14:26] Now, because it's not working right now,
[14:28] everyone's not involved cuz they like
[14:29] narratives on things that they can buy.
[14:31] Same thing with Bitcoin. Same thing with
[14:32] the whole crypto market. It's the reason
[14:34] why last week we talked about Lily. I
[14:35] know when to bring things to people.
[14:37] It's like, okay, they like to buy 52-
[14:38] week high, so I'll just focus on it.
[14:40] Nobody wants to buy Bitcoin right now
[14:41] because it's way below the 200 day
[14:42] moving average. And I get it. Um,
[14:44] narratives are things that give human
[14:46] beings the ability to feel comfortable
[14:48] that they can ride something because the
[14:50] dream is there. A lot of what Elon talks
[14:52] about is a dream, which I believe in,
[14:54] but I also understand from my father the
[14:56] distribution of possibilities of him
[14:59] accomplishing everything he needs for
[15:01] SpaceX to be worth what it is today.
[15:04] It's not I mean, it's a it's he needs to
[15:07] accomplish a lot and so it's a low
[15:08] probability event. I'll take silver and
[15:10] building a position here in silver and
[15:12] in copper. I have copper stuff. I'm
[15:14] buying energy because all of these
[15:16] things that you mentioned to get to the
[15:19] point where we're actually producing
[15:20] compute, we need to have these physical
[15:23] commodities. And what the guy said at
[15:25] all in is I don't know how we're going
[15:26] to get them. It's a very hard time to
[15:28] get them.
[15:29] >> Does that just mean that we're not going
[15:31] to be able to solve the problem and so
[15:32] price explodes or does that mean that we
[15:34] can't solve it in the short term? price
[15:36] starts to go up, but then there's a
[15:37] belief that innovation, entrepreneurs,
[15:39] more investment will then kind of unlock
[15:42] the bottleneck.
[15:43] >> Well, not so you I don't care how good
[15:46] innovation is, you can't get copper out
[15:48] of the ground quickly. You can be the
[15:49] smartest person in the world. If you
[15:51] want to create an algorithm, yeah, you
[15:53] can let that happen. Um, he brought up a
[15:55] point which, and I'm going to I'm going
[15:57] to butcher the name on this, but you
[15:58] guys can all go look it up. Indium, I
[16:01] think it's indium phosphate is a
[16:03] critical mineral. Um, it's very
[16:06] >> that sounds like something I should own,
[16:08] right?
[16:09] >> Well, here's the thing. It's it's a part
[16:10] of the world for Marll. It's a part of
[16:12] the world for Coherent. It's a part of
[16:14] the world for Lumenum. And guess who
[16:16] owns all of it? China.
[16:19] >> So, one of the reasons that I fully
[16:22] believe that the straight of Hormuz will
[16:25] open up when it needs to is China needs
[16:27] oil. China needs to import. Most of the
[16:30] oil they get are from Venezuela and Iran
[16:33] and Russia. And so when you go through
[16:34] it, we've taken away Venezuela and we've
[16:36] gone through it. So magically the
[16:38] coherent CEO was flying over to Trump
[16:40] when he went to go visit shei. And if
[16:43] you go look up indium indium phosphate,
[16:45] you'll realize it is the critical
[16:47] mineral in all data centers. You can't
[16:49] build data centers without it. And they
[16:51] control it. And they've kind of shut it
[16:52] off like they did rare earth. So when
[16:55] you get into these games behind the
[16:57] reason I bring it up is we need so many
[17:00] physical commodities that are impossible
[17:03] for us to get that if China and the US
[17:05] are continuing on this battle the
[17:07] friction of being able to get all this
[17:09] stuff done on the input costs is really
[17:10] difficult and that's why on the
[17:11] commodity side there will be hoarding we
[17:14] will continue to do it I think the US is
[17:16] trying to hoard energy China's hoarding
[17:18] all of the commod all of the minerals
[17:19] that they spend time on we have to
[17:21] figure a way to make new minerals that's
[17:22] what he was talking about so I just want
[17:24] to make sure when people go through this
[17:25] they realize this is not a smooth path.
[17:27] There are a lot of obstructions to
[17:29] getting to those numbers. And I've been
[17:30] talking about an AI midcycle slowdown,
[17:32] which to me is all about that the growth
[17:34] rate that we had already was a lot of
[17:36] hoarding. And now we're going to be in a
[17:38] phase for the next 3 to 6 months where a
[17:40] lot of these semiconductor names and a
[17:41] lot of the cooling names and a lot of
[17:43] all the things on the infrastructure.
[17:45] We're probably going to have equal news
[17:47] of disappointment with equal news of
[17:50] surprises because we just made it
[17:52] through where we shocked everyone and
[17:54] now we've got expectations of some of
[17:56] the largest earnings growth in the
[17:57] history of the market. So
[18:00] today's episode is brought to you by
[18:02] Figure. We all know the DGEN dilemma.
[18:04] You need cash to live your life. Maybe
[18:06] buy a house, run a business, or buy your
[18:08] morning coffee. But the second you sell,
[18:10] you're hit with a massive tax bill and
[18:12] you lose your seat at the table. That's
[18:13] why I've been looking into Figure.
[18:15] Instead of selling and triggering a
[18:16] taxable event, Figure let you borrow
[18:18] against your Bitcoin. You get the
[18:20] liquidity you need at around 8.91%
[18:22] interest. It's 9.9% APR, but you keep
[18:26] 100% of the upside. Now, I know what
[18:28] you're thinking. Is my collateral safe?
[18:30] Figure uses MPC custody. It means your
[18:33] Bitcoin stays in its own segregated
[18:34] wallet. It isn't being rehypothecated or
[18:37] sat on some shaky exchange susceptible
[18:39] to rug poles or other shadiness. Plus,
[18:42] they've got optional liquidation
[18:43] protection for when the market gets well
[18:45] like the Bitcoin market gets. So, if you
[18:48] want to stop selling your future to pay
[18:49] for your present, check my link in the
[18:51] show notes or go to figurearkets.co/pom.
[18:56] >> All right, you ready for this? Uh, very
[18:58] in line with this. I had to look this up
[18:59] to make sure I don't screw this up.
[19:00] >> Okay,
[19:01] >> so there's a guy Alexander Cortez uh who
[19:04] I've met before. Um, he describes
[19:06] himself as a bro scientist. So he's very
[19:08] into lifting, very into nutrition, very
[19:11] into all of the the things that you
[19:13] would associate with a a bro scientist.
[19:14] Very proud of that fact. One of the
[19:16] things that he obviously is into is
[19:18] creatine.
[19:19] >> In creatine, he realized that there was
[19:22] some ingredients that are not made in
[19:24] America. And so as a pro scientist, if
[19:26] you're into creatine, you might be
[19:28] worried about the national defense of
[19:29] the supply chain of creatine, right?
[19:31] >> Is this the merging of us and Uberman
[19:34] right now? So, so this is uh a true
[19:37] story. So, he starts a company. The
[19:39] company is called Athanor. A T H A N O
[19:43] R. It is described as the specialty
[19:45] chemical platform of American defense
[19:48] industry.
[19:49] >> Now, when as he goes through it, what he
[19:51] basically says is that he believes
[19:52] America shifts towards greater
[19:53] industrial sovereignty. They're
[19:55] rebuilding the underlying chemistry. The
[19:56] mission is American independence in
[19:58] critical nitrogen chemicals
[20:00] >> starting with defense. So what he
[20:02] basically came to and people can go
[20:04] check out the website and stuff, right?
[20:04] But what he basically came to the
[20:05] conclusion of was there was this uh
[20:07] nitrogen type uh chemical that is in
[20:11] creatine
[20:12] >> and he discovered it as a bro scientist
[20:14] and realized hey that's not in America.
[20:15] So he went down this deep rabbit hole
[20:17] but then he realized hey there's a lot
[20:18] of other applications for it. My take
[20:20] away from that is I don't know anything
[20:21] about the science. I don't know if he
[20:23] will be successful or not. Smart guy
[20:24] he's got some funding and and he's going
[20:26] to go try to do this. how many like
[20:28] you're talking about critical uh or rare
[20:32] earths and critical minerals and all.
[20:34] He's talking about chemistry like if you
[20:36] were to make a list of how many things
[20:39] go into the products that we need here
[20:41] in America that are not controlled by
[20:43] America almost on a weekly basis now I
[20:46] realize that list is longer than I
[20:47] thought it was.
[20:49] >> Yeah. And now we're entering the point
[20:50] where we're going to figure out what we
[20:51] need for all this to go through it. So
[20:53] the the figuring out of what we need is
[20:55] going to happen. And remember, if you
[20:56] haven't read the story behind where they
[20:59] found a GLP1,
[21:01] it's a heila monster.
[21:02] >> A what?
[21:03] >> Hila monster.
[21:04] >> What is that?
[21:06] >> Give us a lizard.
[21:08] >> A lizard?
[21:08] >> Yeah.
[21:09] >> It was like in the lizard's body and
[21:10] then they extracted it.
[21:11] >> I And I I again, if you go listen to the
[21:13] interview,
[21:14] >> you're telling me that all these people
[21:14] trying to lose weight are using lizard
[21:16] extract?
[21:16] >> Well, lizards basically don't get
[21:19] hungry, I guess, is what it is. And they
[21:21] figured this out. So they wanted to find
[21:23] out how something could go long periods
[21:26] of time without eating much. I David
[21:28] Ricks and and Jensen Yuang talked about
[21:30] this on their interview from uh David
[21:31] Ricks from Eli Liy, but that is like the
[21:34] beginning part of studying and this is
[21:36] where almost every drug we get where he
[21:38] talks about how they would have
[21:40] scientists that would have to go they
[21:42] would just go every year into various
[21:44] countries and go strip soil and fungi
[21:47] from
[21:48] >> the earth,
[21:49] >> go test it and see what materials or
[21:51] chemistry was in it that they had never
[21:53] seen before and then see if it might be
[21:55] something that could be used on
[21:56] >> Darwin. Wasn't he is he the one who was
[21:58] on the uh the trip around the world on
[22:00] the boat and uh and discovered like by
[22:02] studying a bunch of animals a bunch of
[22:03] like scientific theories?
[22:05] >> Yeah. Andos.
[22:06] >> Yeah. Yeah. Yeah. All right. So, we got
[22:08] Darwin Jensen.
[22:09] >> But this is the way pharmaceutical
[22:10] companies would find new advancements.
[22:12] But again, from the time you'd find it
[22:14] until the time that it went into a human
[22:15] body and it went through it. I mean, it
[22:17] could be a 30, 40, 50 year process. So,
[22:19] the Hila monster thing, it GLP1s had
[22:22] been around for a while, but
[22:24] >> maybe that's where lizard brain came
[22:25] from. You ever heard that term?
[22:27] >> No. But
[22:28] >> yeah, a lizard brain like like you know
[22:30] if somebody's got a lizard brain like a
[22:31] game because they're the ones trying to
[22:32] lose weight. Maybe
[22:33] >> you're teaching me new stuff today.
[22:36] >> All right, let's talk about for a second
[22:38] uh the AI model. So there's these two uh
[22:40] tweets that I saw that I think are very
[22:42] related but but pretty interesting. So
[22:43] the first is that uh Semi analysis the
[22:47] uh research firm went and they did a
[22:49] study. These guys went and they bought
[22:50] subscriptions to I think it was OpenAI
[22:52] and Anthropic and they basically started
[22:54] to use the product and they were trying
[22:55] to measure how many tokens are we using.
[22:57] What is the API cost of those tokens and
[23:00] then under the subscription what are we
[23:01] paying and so are these matching up or
[23:03] not? And the basics were these are being
[23:06] heavily subsidized by the companies. The
[23:08] companies are selling you tokens below
[23:10] their actual cost of production and
[23:12] therefore the companies may be driving
[23:13] tons of revenue but they're actually
[23:14] losing money in the short term. We've
[23:16] seen the story before. We've seen
[23:17] venture capitalists basically subsidize
[23:20] uh ride sharing, many other industries.
[23:23] At the same time, there is a guy who was
[23:26] tweeting saying that he thinks somewhere
[23:28] around 80 85% of queries could be moved
[23:31] off of the latest models. And so you
[23:34] don't need to use Fable 5 to figure out
[23:36] what the weather is or you know what uh
[23:38] the address of something is. Instead,
[23:40] you may be able to use just cheap free
[23:42] uh Chinese open source models.
[23:45] that feels like a lot of pressure on
[23:48] these AI companies. So on one hand you
[23:51] have at some point your investors are
[23:52] like stop subsidizing and get the
[23:54] positive unit economics and you know
[23:56] make sure that these are sustainable
[23:57] businesses. At the same time your
[23:59] customers are essentially saying wait a
[24:01] minute I am spending way too much money
[24:03] and so rather than having token
[24:04] leaderboards people are now talking
[24:06] about token limits for employees and and
[24:08] actually throttling the use of these
[24:10] technologies internally. Do you worry at
[24:13] all about the anthropic and open AIS and
[24:15] the, you know, pending IPOs and all that
[24:17] stuff from these stories or do you feel
[24:19] like the demand for this is just so off
[24:21] the charts that none of that matters?
[24:23] >> Well, I think there's two separate
[24:25] answers. Of course, it matters from a
[24:27] stock perspective. Does it matter for
[24:29] the companies being a bubble? No. Does
[24:32] it matter from the And I'll go through
[24:34] why. Does it matter from the company's
[24:37] um
[24:38] one of them being in trouble? Everyone
[24:40] thinks Open AI is in trouble. I I don't
[24:43] see it that way because we're still in
[24:44] the very early stages of adoption. Um
[24:47] the very early stages. So I think a lot
[24:50] of the issue is the acceleration again.
[24:51] And I'm going to keep saying this again
[24:53] and again and I think people need to,
[24:55] you know, to fully grasp it. If this was
[24:57] just software, there's no shortages or
[25:00] bottlenecks. Part of the issue that
[25:01] comes in on this is you talked about the
[25:04] price is going lower. The other side is
[25:06] the cost of producing the tokens is
[25:09] going up dramatically. And when I mean
[25:11] dramatically, I mean we're talking about
[25:14] four times where it was a year ago and
[25:16] it might get cheaper, you know, over
[25:19] time, but that's not happening yet. And
[25:20] we have bottlenecks. So, can we give
[25:22] them the capacity they need is one
[25:24] problem to be able to sustain the ARR.
[25:26] So, it's great if you have more
[25:27] adoption, but it's like, hey, I we have
[25:30] a thousand people that want to come in
[25:31] the restaurant tonight, but we only have
[25:32] 20 seats. Well, you're not making new
[25:34] seats. So you're in trouble. So you can
[25:36] charge more prices, you make it more
[25:38] exclusive, and that's what ends up
[25:39] happening with great restaurants. The
[25:42] model situation is is going to be a
[25:44] problem in my opinion for the near term.
[25:45] And that's why they're racing to the
[25:47] market. When you connect all the dots,
[25:48] why are they all coming to the market at
[25:50] the same time? Why did Google raise
[25:52] money? Why did Meta say a week or and a
[25:55] half later that they were going to come
[25:56] to the market for tens of billions? Why
[25:58] is SpaceX, which is an AI company,
[26:00] Anthropic, Open AAI, and then Open AI,
[26:03] Wall Street Journal article this week,
[26:05] price competition, we're going to lower
[26:06] prices. Like, I I think it's a it's a
[26:09] problem. And if that was not enough, you
[26:11] mentioned something critical. Deepseek
[26:14] is a major story that did not get any
[26:17] press in the last month. So they not
[26:19] only released their model which on a
[26:22] comparable basis is very close to the
[26:25] models
[26:26] but they reduced their price cost
[26:29] dramatically. the AI uh Chinese stocks,
[26:31] the model stocks
[26:33] are selling off and the reason is
[26:35] because they're in a deflationary
[26:37] situation and that has always been my
[26:39] belief is that these models will be
[26:41] commoditized that you don't need to use
[26:43] the most expensive ones and that the
[26:46] corporations haven't figured out yet how
[26:48] to use them. When you put all that
[26:49] together and then you add in Bernie
[26:51] Sanders and Donald Trump want to own
[26:53] part of the companies. I I posted on X.
[26:55] There are so many headwinds when the
[26:57] Pope speaks out against the models.
[26:59] Like, how many things do you want to
[27:00] have where people are like, "Slow this
[27:02] down." And then Daario released just
[27:05] yesterday. I think it was yesterday,
[27:07] maybe it was this morning, I don't know,
[27:08] but he released and I forget the name of
[27:10] it, but he released why we're at a
[27:13] critical point. Now, again, he does this
[27:15] all the time, but I happen to agree with
[27:18] one thing. We are getting closer and
[27:20] closer to AGI because coding is getting
[27:22] better and better. That is the next step
[27:24] to it. We've been doing it without
[27:26] Blackwell and Ver Rubin. Now we're going
[27:28] to have those as well. This is going to
[27:30] accelerate. So for everyone like just
[27:33] thinking there's no downside and that
[27:35] these curves are going to walk out.
[27:38] It's not going to happen that way.
[27:40] >> Why is everyone going public when the
[27:42] music's on? You better be dancing.
[27:44] George Soros, when I see a bubble, I
[27:46] rush in to push it higher. Like that's
[27:48] why, right? is like you go public now or
[27:50] you may not get a chance in 12 months,
[27:53] 18 months, whatever. So that's first.
[27:56] Second, the number one thing I tweeted
[27:59] this back in May and you know people,
[28:01] they're still trying to figure out if I
[28:02] know what I'm talking about with the AI
[28:03] stuff. I said, every single conversation
[28:07] I'm having with the CEO of private
[28:09] companies right now, the mandate from
[28:11] heaven 12 months ago, everyone use AI.
[28:13] Now, wait a minute, these bills are too
[28:16] high. how do we go and try to figure out
[28:18] some different way doesn't mean we don't
[28:20] want you to use the AI but we need to
[28:22] get smarter about how we efficiently
[28:25] consume these tokens right so it was we
[28:27] want to use AI we just want lower cost
[28:28] and so naturally there there's an
[28:30] economic incentive to go pursue this
[28:32] >> now what I also find really interesting
[28:34] is people in terms of users whether it's
[28:37] open AI anthropic we see it with Sylvia
[28:40] etc they are learning what the
[28:43] limitations of these products are but
[28:45] they're realizing that the limitations
[28:47] are way further than they thought they
[28:49] were. And so you I have the Sylvia data.
[28:53] It used to be that people ask six
[28:54] questions per week per user.
[28:56] >> Mhm.
[28:56] >> It's a lot. Almost one a day. It's now
[28:59] 15.
[29:01] I'd like to think we're some geniuses
[29:02] and we made the user experience so much
[29:04] better. Whatever. No. People started to
[29:06] realize what can I do with this thing?
[29:09] The thing got smarter because we keep
[29:10] upgrading the models and therefore you
[29:12] at from a demand perspective ask more
[29:14] and more questions. That's got to be
[29:16] happening. You're you're telling me that
[29:17] Fable 5, you know, uh, queries per
[29:19] person per week is not higher than
[29:22] whatever, you know, opus 4.6 or what? Of
[29:24] course, it is.
[29:25] >> And so, as you see this going, it again
[29:28] is not just a look at today's demand and
[29:30] look at today's supply. This is a very
[29:32] weird market where demand is actually
[29:35] going exponential
[29:36] >> and the supply is not responding in the
[29:39] way that it needs to. So, the problem
[29:41] you're identifying today is actually
[29:44] getting worse. And I think that's maybe
[29:46] the the aspect that people don't quite
[29:48] yet wrap their head around is if the
[29:50] models are getting better and demand is
[29:54] also exponentially increasing and that's
[29:56] just humans we haven't talked about the
[29:57] AI agents doing all this stuff
[30:00] how does this end like do we get to a
[30:03] point where do you see the ad uh product
[30:06] so when you when you uh put in a query
[30:08] somebody built a plugin
[30:09] >> Mhm. So you put in a query, you got to
[30:11] wait. Well, why don't I just serve you
[30:13] ads and if you watch the ads, then you
[30:15] can get paid to cover some of the cost
[30:17] of the compute that you're using at the
[30:18] same time.
[30:19] >> So like you people are going to start
[30:20] doing what the internet does, right?
[30:23] >> Do we just end up in a world where
[30:25] everyone is watching the spinning dial
[30:26] and just, you know, waiting and waiting
[30:28] and waiting and waiting.
[30:29] >> Well, I I think you brought up a point
[30:31] that that I've said to people when they
[30:33] get too negative. So the one side it's
[30:35] like, okay, don't be too positive. And
[30:37] the other side, you're talking about
[30:38] innovations that are going to happen
[30:39] that they're going to figure out. So I
[30:42] I'll give you two examples. One is when
[30:43] you get on an airplane, there's first
[30:45] class, there's business class, there's
[30:47] comfort, there's all these different
[30:48] things and they're priced differently.
[30:51] Um the amount of capacity that Anthropic
[30:53] has access to. You mentioned that
[30:55] there's a lot of things that are costing
[30:57] them money, but then on the flip side,
[31:00] they have a lot of high margin agentic
[31:02] stuff and for inference. Now, if a
[31:05] cancer company is willing to pay for
[31:07] this cuz they have a product that could
[31:09] make them a ton of money, then why
[31:10] shouldn't they be using this than Tom
[31:12] and Mary using it to just chat and
[31:15] paying the $20 which is costing them
[31:17] money cuz they're using it all the time.
[31:18] Eventually, what'll happen is that'll
[31:21] change and they're going to direct
[31:23] people to cheaper models because they're
[31:24] like, "Hey, it's $5 if you use Opus 4.7,
[31:28] which is a great model. It's just not as
[31:29] good as Fable." and they'll start
[31:32] figuring ways to get that capacity off
[31:34] of there over time. There'll be ways for
[31:36] them to do an open source model. We're
[31:39] starting to see more open source models.
[31:40] So, we didn't talk about it, but Neatron
[31:42] Neatron, which is the Nvidia one, it's
[31:45] pretty high on the list. I was shocked
[31:47] at how well it did. I mean, Nvidia puts
[31:50] out an open source model to compete with
[31:51] the Chinese models, and it's not that
[31:53] far below. I read some of the things on
[31:55] there. I mean, it's it's it's a lesser
[31:57] model, but the question is, do you
[31:59] actually need those? And if you can get
[32:00] a laptop and a phone that has Neatron on
[32:02] it, is that a better way to do things?
[32:04] So, I think we're we're missing that
[32:06] between Yeah. advertisements will be on.
[32:08] You're going to have this. You'll have
[32:10] surge pricing like, "Hey, just like you
[32:12] do with Uber. It's like, hey, right now
[32:14] we've got this much capacity. Who wants
[32:15] it? You can pay the most. This is what
[32:16] you're going to pay for." I think
[32:18] everything is going to be over time
[32:20] automated more. I just think that this
[32:22] period for the next three to four years,
[32:24] we just don't know what's how
[32:26] commoditized it'll get, how difficult
[32:29] it's going to get for all these
[32:30] companies. And I've believed all the
[32:32] time and I still believe it. I don't
[32:33] think all of these companies are going
[32:35] to win in the end.
[32:37] >> Who's going to lose?
[32:39] >> Come on, we're here for Alpha. So I I
[32:41] will say that if you ask me between
[32:44] Daario and Sam Alman
[32:48] who I think now Sam Alman in my guess is
[32:50] not going to be the leader of OpenAI at
[32:52] some point soon. That's just my guess.
[32:54] >> Soon.
[32:55] >> Yeah. I I would be shocked if with
[32:57] within a year if he's still running
[32:59] OpenAI. That's just me.
[33:01] >> Because he doesn't want to run or
[33:02] because they may ask him to step down.
[33:03] >> I I'm going to say both. Um remember the
[33:06] incident at his house and all this
[33:08] stuff. I I I'm I'm not sure it's fun
[33:11] like to be in that position right now.
[33:13] Uh I just think that Daario has chosen a
[33:16] political angle in this whole thing
[33:18] which is fairly popular with half of the
[33:21] country and I think Sam Alman has nobody
[33:24] who seems to really like him. Um but the
[33:28] product people do like and I still use
[33:30] it at this point or not still. I did
[33:32] have a period where I used Claude almost
[33:34] exclusively and it went down
[33:35] significantly but now I've gone back to
[33:37] Chad GBT. So I just think in the end
[33:39] OpenAI has an uphill battle because they
[33:41] made a bad decision. They tried to
[33:44] choose the consumer Apple side as
[33:46] opposed to the enterprise side. And I
[33:48] think as long and since Andre Carpathy
[33:52] was brought into um CL uh Anthropic and
[33:55] I keep hearing that they're just
[33:57] attracting the best talent now. I think
[33:59] it's really going to be hard for them to
[34:00] catch up in something as important as
[34:02] coding which will lead to AGI. I I just
[34:05] think at the end anthropic is is is the
[34:07] winner of that battle from from my
[34:09] perspective of
[34:10] >> I mean they are definitely the winner
[34:11] today right um in growth rate revenue
[34:15] momentum you know all all of that stuff
[34:17] um it is a different market that they're
[34:19] going after but but I do think that
[34:20] that's a key part the big thing I wonder
[34:23] is uh can they go and strike uh open can
[34:26] they go and strike deals like could they
[34:28] be part of the Apple ecosystem and you
[34:31] know somehow get their product embedded
[34:33] I I have no clue But, uh, that seems
[34:34] big. All right, guys. Let's talk about
[34:36] something that's actually moving the
[34:38] needle in crypto right now, and that's
[34:39] my guys over at Arch Public. These guys
[34:41] are killing it. They continue to lead
[34:43] the way in everything that's got to do
[34:45] with Aentic Trading. And everyone knows
[34:46] this. Nothing else is being talked about
[34:48] in the market other than let your AI
[34:50] agents trade whatever asset you want.
[34:52] Now, with the addition of ArchAI, the
[34:54] market wave algorithm, and their tax
[34:56] harvest tool, they have turned this into
[34:58] a true one-stop shop for automated
[34:59] trading. They have now over 25,000
[35:02] users. That includes several corporate
[35:03] treasuries, asset managers, and
[35:05] businesses. And so the testing phase for
[35:07] Arch Public that is over. They have
[35:09] perfected how to automate advanced
[35:11] trading strategies without ever taking
[35:13] custody or asking you for your private
[35:15] keys. It's a big deal obviously. So Arch
[35:17] Public is an action speak louder than
[35:19] words type company. That's what I like
[35:21] to see. They give every user access to
[35:23] their platform absolutely free. You get
[35:25] to test it. You run the strategies. You
[35:27] watch the performance in real time. And
[35:29] then only after all of that do you put
[35:31] real money in to go put it to work when
[35:33] you're 100% comfortable. That's it. No
[35:36] pressure, no hidden fees, no gotchas,
[35:37] none of that nonsense. Go to
[35:39] archpubic.com right now and you can get
[35:41] started for free. Once you see how you
[35:43] can accumulate and manage crypto this
[35:45] way, you'll never trade the same again.
[35:47] archub.com. Go check them out today.
[35:52] Today's episode is brought to you by
[35:54] Simple Mining. Bitcoin mining has a
[35:55] reputation for being complicated, risky,
[35:57] and hard to evaluate as a real
[35:59] investment. If you're considering mining
[36:01] in 2026, what actually matters isn't
[36:03] headline profitability. It's uptime,
[36:05] repairs, and whether the operation is
[36:07] run like a real business. That's why
[36:09] I've been using Simple Mining. They're
[36:11] based in Cedar Falls, Iowa, and they run
[36:13] a white glove hosting operation where
[36:15] you own your miners. You choose your own
[36:17] pool, and you have Bitcoin sent directly
[36:19] to your wallet. They were featured on
[36:21] the Inc. 5,000 list as the fastest
[36:23] growing company in Iowa with over 40,000
[36:25] machines under management. What stands
[36:27] out to me is execution. They have the
[36:29] number one rated ASIC repair center. And
[36:31] for the first 12 months, repairs are
[36:33] included. If mining margins get tight,
[36:35] you can pause with no penalties. And if
[36:37] you want to resize or upgrade your
[36:39] fleet, there's a marketplace to resell
[36:40] equipment instead of being stuck. To
[36:42] help people think it through whether
[36:44] mining actually makes sense right now,
[36:45] they put together a short resource
[36:47] called the 2026 Bitcoin mining
[36:49] blueprint. It walks through the five
[36:51] mistakes investors make when allocating
[36:52] to mining, and they also explain how to
[36:54] avoid them before deploying capital. If
[36:56] it sounds interesting to you, you can
[36:58] get it for free at simplemining.io/p.
[37:02] That's simplemining.io/p.
[37:05] Go check it out today and see if you
[37:06] should get into the mining game.
[37:09] Let's talk um you and I are both Knicks
[37:11] fans.
[37:12] >> We're not going to spend a lot of time
[37:13] on the fact that uh the New York Knicks
[37:15] are going to win the NBA championship
[37:16] and uh you know, greatest comeback of
[37:18] all time, etc. U we could talk all day
[37:20] long about how uh Wimi is the greatest
[37:23] choke artist of the NBA. Um or that he's
[37:26] soft and got hit in the head with an
[37:27] egg. We're not going to talk about all
[37:28] that. I
[37:28] >> guess you're trying to lose the San
[37:29] Antonio audience now.
[37:30] >> Well, you know, Charles Barkley already
[37:32] shared a lot about San Antonio. I'm not
[37:33] going to go there. Um but but uh you
[37:36] went to the game as did I and you had a
[37:38] very interesting insight which the
[37:40] Knicks NBA Finals experience
[37:44] drew something back to your thought
[37:45] process around AI, crypto, etc. explain
[37:48] a little bit uh what your takeaway was.
[37:51] >> Well, first of all, if you go to my
[37:52] YouTube channel and you look at my image
[37:54] there, I have a Knicks hat on. That's
[37:56] from years ago. Um the Knicks are very
[38:00] important part of who I am as a person.
[38:02] I'm not going to go in great detail.
[38:03] I'll probably write something about this
[38:05] soon, but let's just say I I went to
[38:07] hundreds of games uh as a kid myself,
[38:11] meaning my father never took me to a
[38:14] Knicks game ever. I decid my when my
[38:16] parents got divorced, I started going to
[38:18] the Knicks game. And I was a young kid,
[38:20] skinny, acne,
[38:22] introvert, and someone who, let's just
[38:25] say, was more on the popular side of
[38:27] school, uh, kind of adopted me. I was a
[38:29] good athlete, um, captain of some of my
[38:32] teams and, um, had a brain, and he
[38:34] identified that, but I was a massive
[38:35] sports fan and I was a bit of a, uh,
[38:38] strange kid in the fact that I could
[38:39] memorize baseball cards and I knew stats
[38:41] on everything and I was an analytical
[38:43] geek. and we just started talking sports
[38:44] and he and I started going to the Nick
[38:46] games. Uh for those of you old enough,
[38:48] Harry M. Stevens, which used to do the
[38:50] um the refreshments and everything
[38:53] there, selling the cotton candy and the
[38:54] stuff in the stands. Well, let's just
[38:56] say to get in the games, I paid those
[38:58] people. I would show up with my friend.
[39:00] We'd pay them five bucks and they'd
[39:02] bring us up the back halls and bring us
[39:03] into the game. And
[39:04] >> in those days when they were really
[39:06] door,
[39:06] >> yeah, they were really bad. Uh I'd sit
[39:10] wherever I wanted cuz there's nobody in
[39:11] there. Uh the scalpers, I knew all of
[39:13] them by name. We would sit outside and
[39:15] they'd throw the tickets, crumbled
[39:16] tickets at us because we would wait
[39:17] until the end. We said we'd give them
[39:19] five bucks. They would hold out and then
[39:20] at halftime they'd give us the tickets
[39:21] we'd go in. So it was a huge part of me
[39:24] starting to take risk. It was a huge
[39:25] thing. But I remember all of this and in
[39:27] 1999 after having gone to most of the
[39:30] important games because luckily I
[39:32] started making enough money to be able
[39:34] to afford them. My friend and I went in
[39:36] 99 and that game ends. Disappointment.
[39:39] We're depressed. We're both Nick fans.
[39:41] were both Met fans and Ranger fans and
[39:44] Jet Well, I was a Jet fan. He was a
[39:46] Giant fan, but next year we go to the
[39:47] Subway Series and here's where the long
[39:49] story ends. Um,
[39:51] >> other than your poor team selection,
[39:52] other than the Knicks.
[39:53] >> Yeah. No, no, no. Let's just take it for
[39:54] where it is. So, we go to the Subway
[39:56] Series games and at the end of them
[39:58] losing, he turns me cuz I bought all of
[40:00] the tickets to all of the games and he's
[40:01] like, "Are we going to have enough money
[40:05] to buy tickets when these teams finally
[40:08] win?"
[40:09] So he died a year later in 911.
[40:12] And so the hard part for me in going to
[40:14] hundreds of games with that one person,
[40:17] how do I go back to a game? And the
[40:18] Knicks are my team. Like that is it. The
[40:20] other teams I I like. Um the Knicks I
[40:23] love and it's it's a passion for me. So
[40:24] it took me a while to go back to the
[40:26] Garden. So fast forward to this
[40:28] particular event.
[40:30] >> Um Wednesday was Wednesday the the
[40:32] >> greatest basketball game ever played.
[40:34] >> The g greatest comeback ever. And I will
[40:37] say um I've become good friends and
[40:39] actually advisor to a company Candy
[40:42] Digital and we both know Tad Smith who's
[40:44] involved with it and I got to to go with
[40:46] them. Now Candy Digital the reason I got
[40:48] interested in what they were doing is
[40:50] because of the experiential side. So,
[40:52] I've written about my belief in stable
[40:54] coins, tokenization, and NFTTS as the
[40:56] three stools of the utilization and the
[40:58] network effects of crypto. And being at
[41:01] the game Wednesday, uh, as a sports fan,
[41:04] when you win two in a row on the other
[41:06] team's home court, your brain is
[41:07] immediately, okay, we got a 90 plus%
[41:09] chance of winning this thing.
[41:11] >> Knicks fans were outside with brooms.
[41:12] >> We're we're we're gonna have this done.
[41:14] Then you go to game three, you watch it,
[41:17] and you're like, okay, you know, we were
[41:20] a little too high on things. We had to
[41:21] come when you're in the stadium and this
[41:24] is what I thought about in game four. So
[41:25] for crypto people, this is what I
[41:27] believe. The world is filled with AI
[41:30] slop and it's only going to get worse.
[41:32] The world is fake filled with deep
[41:33] seats, deep fakes. It's only going to
[41:35] get worse. This will never stop.
[41:38] Anything that you watch, there will be
[41:39] billions of movies. There will be it
[41:41] will be impossible. There will be
[41:42] billions of podcasts and there will be
[41:45] impossible to find stuff. The best
[41:47] content will rise to the surface. But
[41:49] the experiential side of Wednesday night
[41:51] blew me away. So I sat there thinking
[41:53] about NFTTS from the community aspect of
[41:56] being there. It was a game where I
[41:58] thought they were halftime. I'm like,
[42:00] "Oh my god, my brain, the anxiety that's
[42:02] happening. We not only won the first
[42:04] two, but now we're going to lose the
[42:05] next two, have to go back. We lost
[42:07] homecourt advantage." So I was already
[42:09] extrapolating if they lose, how bad
[42:10] it'll be. You can't replace that with
[42:12] AI. That feeling when I left and I
[42:15] headed back to Brooklyn, I took subways.
[42:19] Everyone on there was screaming. It was
[42:22] filled at
[42:23] >> the subway the subway employees.
[42:26] >> They were honking the horn when they
[42:29] came into the the station at both
[42:30] places. So the city itself, the
[42:34] community, and these were people not at
[42:35] the game. So people realize
[42:37] >> the community aspect, the experiential
[42:39] side of I was there, I'm a Nick fan, all
[42:42] of those things. You guys don't have to
[42:43] believe in NFTTS because of the art side
[42:45] and all of the things you saw, but the
[42:47] true aspect of how important that event
[42:49] was and how it cannot be replaced by me
[42:51] by a TV show, by anything. And being
[42:54] there and feeling it was the most
[42:55] powerful thing. And so in a world of AI,
[42:58] in a world where you don't know what's
[43:00] real on anything, I fundamentally
[43:02] believe the blockchain and crypto is
[43:04] critical to all this stuff. So, it was a
[43:06] big event for me on many, many levels,
[43:08] but it was especially a big event to
[43:10] connect it back to this thing of having
[43:13] an NFT for it, to knowing that I'm part
[43:15] of the Knicks community, feeling it
[43:16] afterwards, and just feeling the special
[43:18] nature of what the city was like.
[43:20] >> All right. I I uh I won't play devil's
[43:22] advocate for this.
[43:22] >> Okay.
[43:23] >> Because uh while you were talking, I
[43:25] remembered the tickets I had were on my
[43:28] phone,
[43:28] >> you know, Apple wallet, I go up, very
[43:31] easy to get in, by the way. Anything you
[43:32] read on uh uh in the media that it was
[43:34] hard to get in was not my experience. I
[43:36] don't know about you.
[43:36] >> Ticket Master is an early edition of
[43:38] NFTs.
[43:39] >> Yeah. So those tickets though on your
[43:42] Apple wallet disappear in the sense of
[43:44] you can't do anything with them.
[43:45] >> Mhm.
[43:46] >> So what did the Knicks do? They gave out
[43:48] every fan who showed up got three
[43:50] things. You got a t-shirt.
[43:52] >> Monday night t-shirts were awesome. Like
[43:55] old school wrestling t-shirts basically.
[43:57] get a towel and then they handed you a
[44:01] physical inside of a laminate case, a
[44:04] physical commemorative ticket. And I
[44:07] said to myself, it's like hilarious that
[44:11] everything went digital, but people
[44:13] still want the commemorative ticket.
[44:16] Now, the reason why that's interesting
[44:17] to me is Darren Rovel, sports business
[44:20] guy, super into collectibles and
[44:21] physical stuff. Um, couple hot takes
[44:24] here and there, doesn't like Bitcoin,
[44:25] but that's besides the point. Um
[44:28] he was talking about those commemorative
[44:30] tickets from the game Wednesday night,
[44:32] game four, greatest comeback of all time
[44:35] already by Thursday morning like 50 of
[44:38] them had sold on eBay. And so you say to
[44:41] yourself, wait a second, this isn't even
[44:42] a real ticket to the game. This is
[44:43] basically like a a chachki that got
[44:45] handed out right now. It has a story
[44:47] behind it. It has, you know, in the
[44:49] moment like the morning after versus
[44:51] will they be valuable in 10, 20 years?
[44:53] Who knows, right?
[44:55] But isn't them giving that commemorative
[44:57] ticket, the fact that people watch
[44:58] something physical as the like memory?
[45:01] >> No.
[45:01] >> Or do you think it's a a both? It's not
[45:03] an eitheror.
[45:05] >> So, let's put it this way. Um,
[45:08] tomorrow, if these things are selling on
[45:10] eBay for any amount of money, there'll
[45:13] be a flood of fakes in 2 seconds made,
[45:15] and there'll be 24 million of them sold
[45:18] at whatever price people want to pay for
[45:19] until the price drops. This is the whole
[45:21] point of scarcity. This is the whole
[45:23] point of is it a real thing or is it a
[45:25] fake thing? Well, you don't know unless
[45:26] it's on the blockchain. And this is my
[45:28] belief on on on AI and deep fakes. This
[45:31] is my belief in everything is that
[45:32] people are underestimating that that
[45:35] event is much better for two reasons on
[45:39] the blockchain. Number one, it's
[45:41] actually proof that you were at the
[45:42] game. Number two,
[45:45] you don't have to worry about losing it.
[45:48] You have it. It's in your wallet. It's
[45:50] part of your DNA. It's a passport. It is
[45:52] what it is. This is the whole point of
[45:55] going through it. So, I just I sent a
[45:57] photo of the one that I got at game one
[45:59] at game three and then sent it out at
[46:02] game four. But that thing means nothing
[46:04] to me. And it comes from a person who
[46:06] handed down to his son Bernard King
[46:08] jerseys. I handed down to him the New
[46:11] York Post headline of the Subway series
[46:13] when they were starting it with all of
[46:15] the tickets to the game, the physical
[46:16] tickets to the game. When my friend died
[46:18] in 911, the obituary in the New York
[46:21] Times said that in his wall were Nick
[46:23] tickets and Bruce Springsteen tickets
[46:24] everywhere. So again, those physical
[46:26] tickets
[46:28] >> that's been in the past.
[46:29] >> Did you say Did you save yours?
[46:31] >> Save which one?
[46:32] >> The commemorative ticket from Game Four.
[46:34] I'm I'm about to break your heart or
[46:35] make you very happy.
[46:36] >> Yes, I did.
[46:37] >> Okay. I I I have one. Thankfully, my
[46:39] wife also uh saved hers right now on
[46:42] eBay.
[46:42] >> Yep.
[46:42] >> They're going for between $300
[46:45] >> Mhm. And I see some listed as high as uh
[46:48] 450.
[46:50] I mean that that that's crazy. By the
[46:52] way, that's more the commemorative
[46:54] tickets from game four might be more
[46:55] than some of the Spurs tickets were
[46:58] >> earlier in the playoffs.
[47:00] >> Yeah. Again, I I think it's very
[47:01] difficult for people to envision the
[47:03] world that's coming due to AI. But
[47:06] again, I do this as a joke, but I'm
[47:08] like, if you're not thinking that five,
[47:10] if you go back five years,
[47:13] that is after co, that's how soon five
[47:16] years ago was. Well, 5 years from now,
[47:18] we're going to have humanoids
[47:20] >> in the street. We're going to have
[47:21] humanoids in many, many places.
[47:24] >> So, you can't envision the world that is
[47:26] going to be here when everything you
[47:28] don't know if it's real or fake. You
[47:30] already really don't know that with
[47:31] anything. And people might think they
[47:33] are, but I'm telling you things will
[47:35] have to be put on the blockchain. And
[47:37] whether it's real world assets that are
[47:38] converted, my belief, and that's why
[47:41] these things right now might have some
[47:42] value to people because they want to
[47:44] have them and say that they were there,
[47:45] but the reality is it's not going to
[47:48] >> last thing. Uh we did not plan to talk
[47:49] about this. We only got a couple minutes
[47:50] before we got to go, but uh Jeff Bezos
[47:53] announced Prometheus, I think is how you
[47:55] pronounce it. Um maybe I'm butchering
[47:57] it, but matters. Jeff Bezos's new
[47:59] company. He's co-CEO, which uh he seems
[48:02] pretty involved. They raised something
[48:04] like $12 billion at a $40 billion
[48:06] valuation. Some crazy number. Uh to be
[48:08] honest, like Jeff Bezos, I'm surprised
[48:09] he doesn't just raise at a trillion.
[48:11] Like he's like, "I'm Jeff Bezos and I'm
[48:12] doing something in manufacturing." But
[48:14] the entire idea is he basically wants to
[48:16] use AI and he wants to use hardware to
[48:18] get better at building things. So his
[48:21] thing is if it would take I think the
[48:22] quote I saw was something like 10,000
[48:24] people 10 years to do something, why
[48:25] can't we do it with 100 people in a
[48:27] year, right? How do you significantly
[48:28] increase the efficiency? Um, excited
[48:32] about this? Think this is, you know,
[48:34] Bezos retirement plan and he's just, you
[48:37] know, playing around. What What are your
[48:39] take?
[48:40] >> I I mean, to a degree, this is kind of
[48:43] Chimath's 8020 business. I think it's
[48:46] 8020 or is it 9010? I can't even
[48:48] remember. But regardless, um, let me do
[48:51] 90% of the work with 10% of the people
[48:52] or 80% of the work with 20% of the
[48:54] people. I think when I pay attention to
[48:56] the pharmaceutical industry and I think
[48:58] about how how they're going to move from
[49:01] getting things from phase one and as I
[49:04] mentioned with the Eli Lilly paper when
[49:06] a company that has historically showed
[49:09] the world what it's doing in phase one
[49:12] and now it won't do that because it
[49:13] believes that the ability to catch up
[49:15] due to AI on something like
[49:17] pharmaceuticals is quick that it only
[49:18] makes sense that they would start doing
[49:20] this in the manufacturing and the
[49:21] physical side particularly as we're
[49:23] getting closer to humanoids. So does it
[49:25] make sense? Of course it makes sense.
[49:27] And I think this is the trend. I think
[49:28] from the perspective what he's doing,
[49:30] he's trying to get ahead of the curve
[49:31] and make sure that on the physical side
[49:33] there's a company that can be used to to
[49:36] be used in the process. And I think
[49:39] that's where this is all going is that
[49:40] it's much easier to have an outside
[49:43] place be hired to do that as opposed to
[49:46] you figuring out as a company how to
[49:48] spend the money and go through it. And
[49:50] so all of these moving pieces with very
[49:52] smart people in Silicon Valley and this
[49:54] is I think where the MIT grads are going
[49:56] is everything is now towards the
[49:57] physical hardware side because that is
[49:59] the bottleneck. So the one good thing
[50:02] about innovation in human beings if
[50:06] there's a bottleneck that means there's
[50:08] inflation. If there's inflation that
[50:10] means there's margin that can be made
[50:12] and there's a demand for it. Because if
[50:15] inflation at a time where we have
[50:17] constant deflation due to innovation,
[50:19] the way that we solve that problem is we
[50:20] throw a lot of money at it and we throw
[50:22] a lot of in um and a lot of innovation
[50:24] at it. And so I think he's just hitting
[50:26] on a topic that I talk about all the
[50:28] time here, which is the bottlenecks are
[50:29] real and he's trying to get a company
[50:31] that is to help with the bottlenecks
[50:32] regardless of where it is.
[50:33] >> All right, before I let you go, this is
[50:34] my uh one AI use case of the week to
[50:37] blow your mind. We're we're going to
[50:38] start a new series here now that we're
[50:40] in the conversation cathedral. Um Todd
[50:44] Saunders tweeted, "Methos and Fable is
[50:48] unbelievable. I was on a customer call
[50:50] today and had Claude transcribing in the
[50:52] background. As the customer was telling
[50:54] me about the features they wish their
[50:55] current software had, Claude was
[50:57] building the features in real time. By
[51:00] the end of the call, I was able to show
[51:02] a fully working product with the exact
[51:04] workflow they mentioned 15 minutes
[51:05] earlier. Autonomous looped building
[51:08] triggered from a customer call.
[51:12] If you're a salesperson and you can do
[51:14] that, you may become a trillionaire.
[51:17] Like if you can build custom software
[51:20] while on the call with the customer
[51:21] before the end of the call, show them
[51:22] the solution.
[51:23] >> Yeah.
[51:24] >> So, let's just get back to the parents
[51:26] out there of kids. And I met with
[51:28] someone I worked with when I first got
[51:29] to Morgan Stanley. Um he wanted to uh he
[51:33] wanted me to sit down with his oldest
[51:36] kids. So, I I met them um after a
[51:38] workout. I sat down. We talked for an
[51:40] hour and change. And the entire
[51:42] conversation was about artificial
[51:44] intelligence. And you know, I I have
[51:47] this and we talked about this
[51:48] beforehand, but I I put together this
[51:50] this payw wall because there were three
[51:53] components that I think are critical.
[51:54] And this is not a promo for people to go
[51:56] sign up, but this is just the way that I
[51:59] would think about it.
[52:01] So for me, in trying to explain to
[52:04] people what's going on every day, you
[52:06] and I talk about everything that
[52:08] happens. And we don't have to create a
[52:10] script on what we talk about. It's just
[52:12] you and I going through X and reading
[52:13] the same stuff. It's amazing everything
[52:16] that happens in a week. It is impossible
[52:18] for a human being to keep up on that at
[52:20] a job unless they're listening to
[52:22] podcasts where someone is doing it for
[52:23] them. Well, that's what I do on the payw
[52:25] wall is I'm trying to in the weekly
[52:26] video it's give people what's happened
[52:29] for the week. In the payw wall, I go
[52:31] further and I write papers about
[52:33] specific topics and things like that to
[52:35] explain it in a way that hopefully they
[52:36] understand it. That's the signal part
[52:38] like there's value in knowing what's
[52:40] happening and what's going on. The alpha
[52:42] part is for the investors and that part
[52:44] is what names they should focus on. If I
[52:46] talk about Micron, if I talk about
[52:47] Marll, how are they connected? It's
[52:49] giving them the explanation of which
[52:50] theme they fit into in AI and the whole
[52:52] infrastructure. So from an empowerment
[52:54] basis, this is a bull market in AI and
[52:57] there are going to be periods like now
[52:58] where I think we're in a midcycle
[52:59] slowdown. That doesn't mean that it's a
[53:01] bubble. That just means they go
[53:02] sideways. Some of the names have trouble
[53:04] but I think the hyperscalers have more
[53:06] trouble than these guys do. So I try to
[53:07] prevent waves that they can be long and
[53:09] things that can be short what the timing
[53:10] would be. But the final part is the most
[53:12] important and that's what I spent the
[53:14] time with um my friend's son and that's
[53:16] what the parent should do. Your children
[53:18] need to be using AI. So that example you
[53:21] gave
[53:23] anyone can do something like that but
[53:25] they have to use artificial intelligence
[53:28] every day. And so when someone goes,
[53:30] "Well, this is this the major I have and
[53:32] this is the degree, the job I want to
[53:34] get." I'm like, "There are no more
[53:36] silos, guys." Like the silos are over. I
[53:39] know you you think you have to stay in
[53:41] your major. Don't think that way. Take a
[53:43] job, help a company in a different
[53:45] business, help a company in something
[53:47] else. All of those skills, by the end of
[53:49] time, you'll be an AI native person that
[53:51] is able to do the things that you
[53:52] mentioned.
[53:53] >> I um I think that this is just the tip
[53:56] of the iceberg. Like I saw that and I
[53:58] was like, "Oh my god, imagine how you
[53:59] could build things for children, you
[54:02] could build things for customers, you
[54:03] could build whatever." So, um, all
[54:04] right. You're not going to tell people
[54:05] to go sign up. I'm going to tell people
[54:06] to go sign up for your payw wall. Uh,
[54:08] what is the URL for this thing?
[54:10] >> It's, uh, ai.22varche.com.
[54:14] >> All right. AI.22vresearch.com.
[54:17] If Jord's ever helped you, go sign up.
[54:19] Just just do it. I'm not going to give
[54:20] you some hard pitch, whatever. Help the
[54:22] guy out, right? He's doing a lot of
[54:23] work. U, it'd be very nice of you. you
[54:25] know, consider like a father's day gift,
[54:27] maybe a birthday gift, uh anniversary
[54:29] gift, what, whatever.
[54:30] >> And this week I did put up how to build
[54:32] a knowledge brain 30 minutes so you can
[54:34] build Jensen Yuang's knowledge brain
[54:36] that I've talked about on this show.
[54:38] >> Yeah, there you go. So, if you've made
[54:39] money listening in shorty, come on,
[54:40] let's go. It's called a tip. Go tip them
[54:43] at ai.22vresearch.com.
[54:45] Um, thank you for doing this. Do it
[54:46] again next week. Go next.

← Volver al listado de vídeos

Scroll al inicio