La burbuja de la IA

[muyuu]

https://www.anthropic.com/news/series-h

Anthropic raises $65B in Series H funding at $965B post-money valuation


Anthropic has raised $65 billion in Series H funding led by Altimeter Capital, Dragoneer, Greenoaks, and Sequoia Capital, valuing the company at $965 billion post-money.

Global enterprises across industries are deploying Claude in their core operations, and a growing number of people around the world use it for their everyday work. Since our Series G in February, adoption has continued to grow across global enterprise customers, and our run-rate revenue crossed $47 billion earlier this month. This latest funding is expected to advance our safety and interpretability research, expand compute to meet growing demand for Claude, and scale the products and partnerships our customers rely on.

“Claude is increasingly indispensable to our growing global community of customers, and we work tirelessly to make tools like Claude Code and Cowork more helpful, more powerful, and more adaptable to their needs,” said Krishna Rao, Chief Financial Officer of Anthropic. “This funding will help us serve the historic demand we are experiencing, stay at the research frontier, and bring Claude to more of the places where work happens.”

The round was co-led by Capital Group, Coatue, D1 Capital Partners, GIC, ICONIQ, and XN. Significant investors in this round include AMP PBC, Baillie Gifford, Blackstone, Brookfield, D.E. Shaw Ventures, DST Global, Fidelity Management & Research Company, General Catalyst, Insight Partners, Jane Street, Lightspeed Venture Partners, MGX, NTTVC, NX1 Capital, Situational Awareness LP, T. Rowe Price Associates, Inc., T. Rowe Price Investment Management, Inc., and Temasek. It also includes $15 billion of previously committed investments from hyperscalers, including $5 billion from Amazon.

Joining them are strategic infrastructure partners—Micron, Samsung, and SK hynix—whose technologies play a critical role in the world's supply of memory, storage, and logic chips. As demand for Claude continues to grow, these relationships will help us scale our compute reliably at the pace our customers need.

We have significantly expanded our compute capacity in recent weeks. We signed agreements with Amazon for up to five gigawatts of new capacity, with Google and Broadcom for five gigawatts of next-generation TPU capacity, and with SpaceX for access to GPU capacity in Colossus 1 and Colossus 2. Claude is the first frontier model available on all three of the world's largest cloud platforms: Amazon Web Services, Google Cloud, and Microsoft Azure. AWS remains our primary cloud provider and training partner.

“Claude’s latest advancements have driven large-scale adoption among the world’s most demanding organizations. This momentum positions Anthropic to lead the next phase of AI innovation and capture the enormous opportunity ahead,” said Brad Gerstner, Founder and CEO of Altimeter Capital.

“Dragoneer has long partnered with companies building the technology that will shape our future. Anthropic is helping pull forward this future, as intelligence becomes an increasingly critical ingredient to the way businesses operate and how their products show up in the world,” said Marc Stad, Managing Partner at Dragoneer. “The technological progress we are seeing right now is breathtaking. And we believe that we are still in the earliest days of both the development and commercialization of this technology.”

“Anthropic has built an organization in which the world’s best researchers and engineers operate with unmatched clarity of purpose, because they believe this is the most important work they will ever do,” said Neil Mehta, Founder and Managing Partner at Greenoaks. “Rarely has a company’s culture, mission, and commercial momentum reinforced each other so completely. We are honored to deepen our partnership.”

“Startups and Global 5000 companies alike are deploying Claude to handle complex workflows, and in doing so, Claude is learning how businesses actually operate: the context, the processes, the judgment,” said Alfred Lin, Partner at Sequoia Capital. “Anthropic is building the bridge between where enterprise AI stands today and where it's headed.”

We are grateful for the support of our investors and partners as we continue building Claude for people and organizations around the world.

También han sacado Opus 4.8

[pollo]

Una mentira de bulto, que el modelo está aprendiendo no se qué al estar desplegado por las empresas por ahí. El tal Alfred Lin o es imbécil (que es probable) o es un hijo de puta (que es probable).

https://www.anthropic.com/news/series-h

Anthropic raises $65B in Series H funding at $965B post-money valuation


Anthropic has raised $65 billion in Series H funding led by Altimeter Capital, Dragoneer, Greenoaks, and Sequoia Capital, valuing the company at $965 billion post-money.

Global enterprises across industries are deploying Claude in their core operations, and a growing number of people around the world use it for their everyday work. Since our Series G in February, adoption has continued to grow across global enterprise customers, and our run-rate revenue crossed $47 billion earlier this month. This latest funding is expected to advance our safety and interpretability research, expand compute to meet growing demand for Claude, and scale the products and partnerships our customers rely on.

“Claude is increasingly indispensable to our growing global community of customers, and we work tirelessly to make tools like Claude Code and Cowork more helpful, more powerful, and more adaptable to their needs,” said Krishna Rao, Chief Financial Officer of Anthropic. “This funding will help us serve the historic demand we are experiencing, stay at the research frontier, and bring Claude to more of the places where work happens.”

The round was co-led by Capital Group, Coatue, D1 Capital Partners, GIC, ICONIQ, and XN. Significant investors in this round include AMP PBC, Baillie Gifford, Blackstone, Brookfield, D.E. Shaw Ventures, DST Global, Fidelity Management & Research Company, General Catalyst, Insight Partners, Jane Street, Lightspeed Venture Partners, MGX, NTTVC, NX1 Capital, Situational Awareness LP, T. Rowe Price Associates, Inc., T. Rowe Price Investment Management, Inc., and Temasek. It also includes $15 billion of previously committed investments from hyperscalers, including $5 billion from Amazon.

Joining them are strategic infrastructure partners—Micron, Samsung, and SK hynix—whose technologies play a critical role in the world's supply of memory, storage, and logic chips. As demand for Claude continues to grow, these relationships will help us scale our compute reliably at the pace our customers need.

We have significantly expanded our compute capacity in recent weeks. We signed agreements with Amazon for up to five gigawatts of new capacity, with Google and Broadcom for five gigawatts of next-generation TPU capacity, and with SpaceX for access to GPU capacity in Colossus 1 and Colossus 2. Claude is the first frontier model available on all three of the world's largest cloud platforms: Amazon Web Services, Google Cloud, and Microsoft Azure. AWS remains our primary cloud provider and training partner.

“Claude’s latest advancements have driven large-scale adoption among the world’s most demanding organizations. This momentum positions Anthropic to lead the next phase of AI innovation and capture the enormous opportunity ahead,” said Brad Gerstner, Founder and CEO of Altimeter Capital.

“Dragoneer has long partnered with companies building the technology that will shape our future. Anthropic is helping pull forward this future, as intelligence becomes an increasingly critical ingredient to the way businesses operate and how their products show up in the world,” said Marc Stad, Managing Partner at Dragoneer. “The technological progress we are seeing right now is breathtaking. And we believe that we are still in the earliest days of both the development and commercialization of this technology.”

“Anthropic has built an organization in which the world’s best researchers and engineers operate with unmatched clarity of purpose, because they believe this is the most important work they will ever do,” said Neil Mehta, Founder and Managing Partner at Greenoaks. “Rarely has a company’s culture, mission, and commercial momentum reinforced each other so completely. We are honored to deepen our partnership.”

“Startups and Global 5000 companies alike are deploying Claude to handle complex workflows, and in doing so, Claude is learning how businesses actually operate: the context, the processes, the judgment,” said Alfred Lin, Partner at Sequoia Capital. “Anthropic is building the bridge between where enterprise AI stands today and where it's headed.”

We are grateful for the support of our investors and partners as we continue building Claude for people and organizations around the world.

También han sacado Opus 4.8

[muyuu]

https://www.economist.com/finance-and-economics/2026/06/01/can-the-stockmarket-swallow-anthropic-spacex-and-openai
 ( https://archive.is/WNrzP )

Finance & economics | Giga-IPOs

Can the stockmarket swallow Anthropic, SpaceX and OpenAI?


Watch out for indigestion


Listen to this story
AI Narrated


They promise to be the biggest stockmarket debuts ever. On June 11th SpaceX reportedly hopes to raise $75bn from investors, by issuing shares that will begin trading on the Nasdaq exchange the next day. Elon Musk’s rocketry firm will probably soon be followed by two other mammoth listings. Anthropic, an artificial-intelligence lab, filed draft paperwork for its initial public offering on June 1st; OpenAI, a competitor, is expected to do so soon. The two are rumoured to be seeking as much as $60bn apiece. Together, the three giga-IPOs may add as much as $4trn to the market value of listed American companies in a matter of months.

How on Earth will the stockmarket handle this? Headlines predict a “trading frenzy”. Steve Sosnik, chief strategist at Interactive Brokers, one of the world’s biggest online trading platforms, has warned of the “existential risk” the listings pose. A particular worry is that compilers of stockmarket indices will grant the gigantic trio fast-track entry to their benchmarks. That would prompt tracker funds with trillions of dollars in assets to buy the newly minted shares days after they are issued. After exhausting a big pool of buyers straight away, who will be left?
The answer is: lots of investors in an extraordinarily deep and liquid market. Unprecedented as the serving of supersized IPOs is, America’s extraordinary stockmarket will gulp it down. In the years to follow, though, expect some indigestion.

First put the giga-IPOs’ size in context. In nominal terms, the current record for capital raised by a debut listing is held by Saudi Aramco, which in 2019 garnered $29bn ($38bn in today’s money) when it floated in Riyadh. SpaceX, Anthropic and OpenAI are collectively targeting $200bn-odd. Yet this is a rounding error in America’s stockmarket. Firms in the broad Russell 3000 share index have a total market value of $79trn; those in the narrower (but more widely tracked) S&P 500 index of big companies are worth $69trn.

As a consequence, investors in index funds will not immediately see their portfolios change much. Although Nasdaq has already shortened the “seasoning” period before index inclusion to 15 trading days and FTSE Russell has slashed its waiting time to five days (and S&P Dow Jones is reportedly considering something similar), most share indices weight firms in proportion to the value only of shares they have released for public trading (the “free float”). For SpaceX, this means just the $75bn or so of stock it intends to issue in June—so its initial weight in the S&P 500 will be around 0.1%. The NASDAQ 100 is an exception, and has changed its rules to weight companies at up to three times their free float, in an apparent effort to woo Mr Musk. Even so, SpaceX’s probable initial weight in this $40trn index will still only be around 0.5%.



This will change as more shares are released for trading. All but one of America’s listed tech giants have free floats above 85% (see chart 1); the lowest is that of Meta, which went public in 2012 and has 13% of its shares still owned by Mark Zuckerberg, its founder. At first, “lock-up” provisions in the IPO prospectuses of SpaceX, Anthropic and OpenAI will prevent company insiders and early investors from selling their existing stakes and raising the free float. Over time, however, these will expire and trillions of dollars’ worth of new shares will come to market.

SpaceX plans to release its locked-up shares in a series of tranches. If its IPO issues $75bn of shares, valuing the firm at its hoped-for $1.75trn, the initial free float will be 4%. None of Mr Musk’s stake, which accounts for about half of the remainder, can be sold for 366 days after the IPO. This restriction also applies to some shares held by “certain significant investors”.

Lock-ups on the rest, representing a little under half of SpaceX’s value, will expire more quickly. After its first quarterly report, probably in August or September, insiders can sell 20% of their stakes. They can offload another 10% if the shares are then trading 30% or more above their IPO price. Extra tranches are due for release on set dates after the IPO, and after the second quarterly earnings report (see chart 2).



Insiders do not have to sell their shares, of course. Mr Musk, in particular, may hold on to his—most of which carry outsize voting rights and so cement his control of SpaceX. Similar considerations will apply to shareholders in Anthropic and OpenAI after the labs’ flotations. So the addition of these firms to public markets will unfold over years rather than days.

Gradual does not, however, mean inconsequential. If history is a guide, those who buy the resulting shares stand a good chance of disappointment. Jay Ritter of the University of Florida has studied the post-IPO returns of stocks listed between 1980 and 2024. The average such stock returned 20 percentage points less than the broader market over the three years after its first trading day. Firms valued at over 40 times their revenue underperformed by 58 percentage points. SpaceX, with a valuation of $1.75trn, would begin trading at over 90 times its revenue.

Blockbuster IPOs are also often taken as a sign that a bull market is nearing its peak—understandably, since firms want to sell shares for top dollar. The last surge in listings, in 2020 and 2021, came just before a bear market. Previous IPO booms, for instance in the late 1990s or the years before 2008, were followed by far bigger slumps.

Today, if the giga-trio underperforms, it may even precipitate a correction. All three firms are closely associated with progress in AI and so too, increasingly, is the wider market: America’s ten biggest listed AI-related firms already account for two-fifths of the S&P 500’s value. Bad news for SpaceX alone might not harm a tracker fund much; bad news for AI certainly would. Funds that weight each of an index’s component stocks equally, rather than by market value, offer some protection from this. But right now they amount to betting against the market—the opposite of passive investment.

A broader worry still is that the giga-IPOs herald more capital-raising, both by the newly listed tech giants and their older peers. For years, notes Victor Haghani of Elm Wealth, an investment firm, capital has been abundant and shares increasingly scarce. Tech giants have generated so much cash that they have been buying back shares rather than issuing new ones even as white-collar workers have ploughed retirement savings into the market. This drove up share prices.

Now the tech behemoths are slowing share buy-backs or halting them altogether, instead reinvesting their profits to develop AI. Several have turned to the bond market for more capital. And the club’s newest members are tapping the stockmarket. At the same time, white-collar workers are perhaps most in danger of seeing AI automate their jobs and immiserate their pension pots. Expect investors will go ga-ga for the giga-IPOs. A few years from now, though, the stockmarket may need to prepare for a capital diet. ■

[pollo]

https://www.economist.com/finance-and-economics/2026/06/01/can-the-stockmarket-swallow-anthropic-spacex-and-openai
 ( https://archive.is/WNrzP )

Finance & economics | Giga-IPOs

Can the stockmarket swallow Anthropic, SpaceX and OpenAI?


Watch out for indigestion


Listen to this story
AI Narrated


They promise to be the biggest stockmarket debuts ever. On June 11th SpaceX reportedly hopes to raise $75bn from investors, by issuing shares that will begin trading on the Nasdaq exchange the next day. Elon Musk’s rocketry firm will probably soon be followed by two other mammoth listings. Anthropic, an artificial-intelligence lab, filed draft paperwork for its initial public offering on June 1st; OpenAI, a competitor, is expected to do so soon. The two are rumoured to be seeking as much as $60bn apiece. Together, the three giga-IPOs may add as much as $4trn to the market value of listed American companies in a matter of months.

How on Earth will the stockmarket handle this? Headlines predict a “trading frenzy”. Steve Sosnik, chief strategist at Interactive Brokers, one of the world’s biggest online trading platforms, has warned of the “existential risk” the listings pose. A particular worry is that compilers of stockmarket indices will grant the gigantic trio fast-track entry to their benchmarks. That would prompt tracker funds with trillions of dollars in assets to buy the newly minted shares days after they are issued. After exhausting a big pool of buyers straight away, who will be left?
The answer is: lots of investors in an extraordinarily deep and liquid market. Unprecedented as the serving of supersized IPOs is, America’s extraordinary stockmarket will gulp it down. In the years to follow, though, expect some indigestion.

First put the giga-IPOs’ size in context. In nominal terms, the current record for capital raised by a debut listing is held by Saudi Aramco, which in 2019 garnered $29bn ($38bn in today’s money) when it floated in Riyadh. SpaceX, Anthropic and OpenAI are collectively targeting $200bn-odd. Yet this is a rounding error in America’s stockmarket. Firms in the broad Russell 3000 share index have a total market value of $79trn; those in the narrower (but more widely tracked) S&P 500 index of big companies are worth $69trn.

As a consequence, investors in index funds will not immediately see their portfolios change much. Although Nasdaq has already shortened the “seasoning” period before index inclusion to 15 trading days and FTSE Russell has slashed its waiting time to five days (and S&P Dow Jones is reportedly considering something similar), most share indices weight firms in proportion to the value only of shares they have released for public trading (the “free float”). For SpaceX, this means just the $75bn or so of stock it intends to issue in June—so its initial weight in the S&P 500 will be around 0.1%. The NASDAQ 100 is an exception, and has changed its rules to weight companies at up to three times their free float, in an apparent effort to woo Mr Musk. Even so, SpaceX’s probable initial weight in this $40trn index will still only be around 0.5%.



This will change as more shares are released for trading. All but one of America’s listed tech giants have free floats above 85% (see chart 1); the lowest is that of Meta, which went public in 2012 and has 13% of its shares still owned by Mark Zuckerberg, its founder. At first, “lock-up” provisions in the IPO prospectuses of SpaceX, Anthropic and OpenAI will prevent company insiders and early investors from selling their existing stakes and raising the free float. Over time, however, these will expire and trillions of dollars’ worth of new shares will come to market.

SpaceX plans to release its locked-up shares in a series of tranches. If its IPO issues $75bn of shares, valuing the firm at its hoped-for $1.75trn, the initial free float will be 4%. None of Mr Musk’s stake, which accounts for about half of the remainder, can be sold for 366 days after the IPO. This restriction also applies to some shares held by “certain significant investors”.

Lock-ups on the rest, representing a little under half of SpaceX’s value, will expire more quickly. After its first quarterly report, probably in August or September, insiders can sell 20% of their stakes. They can offload another 10% if the shares are then trading 30% or more above their IPO price. Extra tranches are due for release on set dates after the IPO, and after the second quarterly earnings report (see chart 2).



Insiders do not have to sell their shares, of course. Mr Musk, in particular, may hold on to his—most of which carry outsize voting rights and so cement his control of SpaceX. Similar considerations will apply to shareholders in Anthropic and OpenAI after the labs’ flotations. So the addition of these firms to public markets will unfold over years rather than days.

Gradual does not, however, mean inconsequential. If history is a guide, those who buy the resulting shares stand a good chance of disappointment. Jay Ritter of the University of Florida has studied the post-IPO returns of stocks listed between 1980 and 2024. The average such stock returned 20 percentage points less than the broader market over the three years after its first trading day. Firms valued at over 40 times their revenue underperformed by 58 percentage points. SpaceX, with a valuation of $1.75trn, would begin trading at over 90 times its revenue.

Blockbuster IPOs are also often taken as a sign that a bull market is nearing its peak—understandably, since firms want to sell shares for top dollar. The last surge in listings, in 2020 and 2021, came just before a bear market. Previous IPO booms, for instance in the late 1990s or the years before 2008, were followed by far bigger slumps.

Today, if the giga-trio underperforms, it may even precipitate a correction. All three firms are closely associated with progress in AI and so too, increasingly, is the wider market: America’s ten biggest listed AI-related firms already account for two-fifths of the S&P 500’s value. Bad news for SpaceX alone might not harm a tracker fund much; bad news for AI certainly would. Funds that weight each of an index’s component stocks equally, rather than by market value, offer some protection from this. But right now they amount to betting against the market—the opposite of passive investment.

A broader worry still is that the giga-IPOs herald more capital-raising, both by the newly listed tech giants and their older peers. For years, notes Victor Haghani of Elm Wealth, an investment firm, capital has been abundant and shares increasingly scarce. Tech giants have generated so much cash that they have been buying back shares rather than issuing new ones even as white-collar workers have ploughed retirement savings into the market. This drove up share prices.

Now the tech behemoths are slowing share buy-backs or halting them altogether, instead reinvesting their profits to develop AI. Several have turned to the bond market for more capital. And the club’s newest members are tapping the stockmarket. At the same time, white-collar workers are perhaps most in danger of seeing AI automate their jobs and immiserate their pension pots. Expect investors will go ga-ga for the giga-IPOs. A few years from now, though, the stockmarket may need to prepare for a capital diet. ■

Lo que yo me puedo oler es que en el momento en el que estas peten (lo cual ocurrirá porque no hay nada que justifique la burrada que se está poniendo en ellas, monten los numeritos que monten) occidente estará totalmente vendido del todo y quedará en evidencia como la mierda que realmente es bajo la fachada.

Y que conste que esto lo digo porque estoy harto de ver la degradación de los valores y la sociedad en la que vivimos, y a dónde nos ha llevado la religión anglo de la pasta y el clasismo a costa de todo.

[pollo]

Pues nada, lo que iba a acabar pasando.

A ver si ahora sale a cuenta.

https://news.slashdot.org/story/26/06/02/0512209/github-copilot-users-react-to-new-usage-based-pricing-system

GitHub Copilot Users React To New Usage-Based Pricing System

An anonymous reader quotes a report from Ars Technica:
In April, GitHub announced that it was moving subscribers from request-based billing to a usage-based model for its AI-powered Copilot service. As that new pricing model goes into effect today, many GitHub Copilot users are reporting some extreme sticker shock as they realize just how quickly their previous "normal" usage is burning through their newly limited monthly allotment of AI credits. Across social media and forums, many Copilot users are sharing personal statistics showing how just a few hours of AI usage can now account for a large chunk of their new monthly subscription caps. For some users, it reportedly took less than a day to use up a month's usage quota.

That's a big change from previous months, when GitHub Copilot subscribers were allocated a certain number of "requests" and "premium requests" based on their payment tier. GitHub said that the old system meant that "a quick chat question and a multi-hour autonomous coding session [could] cost the user the same amount," forcing Copilot itself to "absorb much of the escalating inference cost behind that usage." [...] Indeed, some Copilot users have been sharing estimates from GitHub's own tool showing that their previous monthly usage would rack up bills in the thousands of dollars under the new pricing plan. Under GitHub's new usage-based pricing system, paid Copilot subscriptions instead grant users a certain number of AI "credits" each month, with one credit corresponding to $0.01 of usage. Subscribers also get bonus credits depending on their subscription level: the $10/month Pro plan includes 1,500 credits ($15 worth); the $39 Pro+ plan includes 7,000 credits ($70 worth); and the $100/month Copilot Max plan includes 20,000 credits ($200 worth).

The precise number of Copilot credits used by a given prompt is determined by the number of input and output tokens used and the rates charged by the underlying large language model. That means pricing is highly dependent not just on the type of request but on the specific model that a user chooses. One million output tokens from OpenAI's GPT-5.4 nano would run just $1.25 on GitHub Copilot, but that same level of output would run $30 on the frontier GPT-5.5 model (Copilot users who rely on "Auto" mode to pick the most appropriate available model for any request should be extremely careful, as some users report it can switch to expensive models for extremely simple queries).

[muyuu]

https://www.expansion.com/economia/financial-times/2026/05/29/6a198dc3e5fdeaa0318b4596.html

https://www.expansion.com/economia-digital/companias/2026/05/29/6a1957ea468aeb7d1a8b4572.html

Saludos.

[muyuu]

https://www.tomshardware.com/pc-components/ddr5/32gb-of-ddr5-now-costs-usd375-minimum-ai-shortage-continues-to-squeeze-pc-building

32GB of DDR5 now costs $375 minimum — AI shortage continues to squeeze PC building
Lower-priced kits are disappearing by the day


As the demands of AI continue to consume manufacturing capacity at every level of the PC hardware supply chain, 32GB of DDR5 RAM — broadly understood to be the sweet spot for gaming PCs and enthusiast builds — can no longer be found for less than $375. Well, $374.97 to be precise.

RAM price tracking through 2026 will show you that kits that routinely cost less than $100 just a year ago are now fetching upwards of $240 (16GB). As the AI frenzy has taken hold, retailers far and wide have been pumping up their RAM prices to exorbitant levels. However, there's so much fluctuation and noise that average pricing is now something of a ludicrous fugazi. The going rate for 32GB of DDR5 RAM — the cheapest you can expect to pay — has hovered around $320 for some time, climbing past $350 in recent weeks. Price tracking courtesy of PCPartPicker now reveals the cheapest 32GB DDR5 RAM you can buy is $375. Specifically, four XPOWER kits from Silicon Power will set you back $374.97 thanks to a promo code. You can see the listings yourself below.


As you can imagine, this is enormous pricing pressure for enthusiasts trying to build gaming PCs or upgrade their rigs in 2026. A component that once cost less than $100 and was something of an afterthought now costs almost four times as much, and that's before you've even fired a neuron in consideration of aesthetics, timings, or brand. More popular kits from the likes of Corsair and Crucial, or RGB offerings to match the rest of your build, will easily set you back more than $400.

Of course, 32GB is really the minimum sweet spot you should be aiming for when building a PC in 2026. If you did want more capacity, 64GB will set you back an astonishing $679.99. 16GB of RAM as a compromise can be found for $200 at B&H Photo, but with SK hynix warning that manufacturing constraints will persist through 2030, there's no sign of prices letting up so that you can upgrade capacity any time soon.

The humble RAM combo deals we've been highlighting in recent months are a small source of solace for builders, letting you score RAM for less than the $375 going rate if you pair it with a decent motherboard, a processor, or even an entire set of PC components. A theme of ongoing Computex 2026 announcements remains a lack of pricing clarity on lots of PC hardware, including Nvidia's RTX Spark laptops and PCs, as well as new-build systems and, of course, RAM components themselves. Vendors are likely wary of scaring off potential buyers with higher-than-expected prices ahead of release. Perhaps more likely, the prices haven't been set because they're still going up. Storage isn't much better, with SSD price tracking revealing that drives which once cost as little as $38 are now fetching $200.

AMD is making a noticeable effort to keep PC gaming prices down, this week announcing the return of its Ryzen 7 5800X3D, and the advent of a new Ryzen 7 7700X3D. Intel, which warned this week that "something has to give" when it comes to memory prices, also teased dragging out some of its legacy products to give users more options on older memory technologies, namely Raptor Lake and DDR4.

seguimiento de precios: https://pcpartpicker.com/trends/price/memory/#ram.ddr5.5600.2x32768


los bots superan a los humanos el tráfico monitorizado por Cloudfare : https://radar.cloudflare.com/traffic#bot-vs-human

la primera vez fue en 2025 y normalmente los fines de semana esto ocurría, pero ya parece que es algo estable

[Cadavre Exquis]

https://www.euribor.com.es/2026/06/02/el-preocupante-entramado-financiero-que-sostiene-la-burbuja-de-la-inteligencia-artificial/

El preocupante entramado financiero que sostiene la burbuja de la inteligencia artificial

Carlos López · 2026.06.02

Cuando Michael Burry habla, Wall Street escucha con atención. El gestor de fondos que se hizo de oro por anticipar la crisis de las hipotecas basura en 2008 ha puesto ahora el foco en el sector tecnológico. Según su análisis, detrás de los ingresos récord de la inteligencia artificial se esconde un entramado de ingeniería financiera que traslada el riesgo de las tecnológicas hacia los fondos de pensiones de los ciudadanos.

Este mecanismo permite que empresas como Nvidia y la filial xAI de Elon Musk muevan miles de millones de euros en equipamiento tecnológico sin reflejar deudas en sus balances, utilizando una estructura legal que recuerda a los complejos productos que provocaron la crisis de 2008.

La creación de empresas pantalla para limpiar el balance

El proceso comienza con una venta de procesadores de última generación por valor de 5.400 millones de dólares. Estos chips son esenciales para entrenar modelos de lenguaje avanzados, pero en lugar de venderse directamente a una gran tecnológica, la operación se realiza a través de una Sociedad de Propósito Específico llamada Valor.

Valor es una empresa fantasma sin empleados ni actividad propia, diseñada exclusivamente para figurar como propietaria legal de los componentes. Los chips se envían directamente a los centros de datos de xAI, la empresa de Elon Musk, que los utiliza en régimen de alquiler mediante un contrato de arrendamiento neto triple. De este modo, la tecnológica asume los costes de mantenimiento y los impuestos, pero evita registrar la deuda de la compra en sus libros contables.

Por su parte, Nvidia registra la operación como una venta cerrada, lo que impulsa sus ingresos trimestrales ante los inversores. Sin embargo, el análisis revela que Nvidia financió parte de la operación aportando 1.900 millones de dólares a la propia empresa pantalla. En el sector, esta práctica se conoce como financiación circular: aportar fondos a tu propio comprador para inflar las cifras de ventas.

El papel del crédito privado en la sombra

Para cubrir los 3.500 millones de dólares restantes, la estructura recurre al mercado de crédito privado a través de Apollo Global Management. Tras las regulaciones bancarias impuestas tras la crisis anterior, las entidades tradicionales evitan financiar operaciones de alto riesgo respaldadas por activos que se deprecian rápidamente. Las firmas de capital riesgo han asumido ese papel, operando con una menor supervisión regulatoria.

Apollo emite títulos de deuda respaldados por los pagos del alquiler de los chips y el valor de los propios procesadores. Posteriormente, vende estos activos a Athene, una aseguradora de su propiedad especializada en la venta de rentas vitalicias para la jubilación.

Los fondos de jubilación como soporte del sector tecnológico

A través de esta conexión, los ahorros de los ciudadanos terminan financiando la adquisición de infraestructura tecnológica de alto riesgo. Los clientes que contratan estos productos buscan estabilidad y asumen que su dinero se invierte en activos seguros, como bonos del Estado. En su lugar, sus fondos sostienen la expansión de servidores informáticos de última generación.

Para evitar que los reguladores nacionales exijan reservas de capital elevadas por mantener activos ilíquidos en cartera, la aseguradora traslada la gestión de estos fondos a filiales radicadas en Bermudas, donde la normativa de solvencia es más flexible.

Activos sin precio de mercado y un apalancamiento elevado

En esta jurisdicción de baja regulación, una tercera parte de la cartera de la aseguradora se compone de activos de Nivel 3. Según las normas contables, estos productos carecen de un precio de mercado observable, por lo que su valoración se calcula mediante modelos matemáticos internos de la propia empresa, un método que en el sector se denomina de forma irónica valoración imaginaria.

El riesgo se incrementa debido a que la estructura opera con un apalancamiento de dieciséis veces su capital. Esto implica que una reducción del 6,25% en el valor real de los procesadores consumiría por completo los fondos propios que sostienen el entramado, comprometiendo la solvencia del sistema.

El impacto de la obsolescencia tecnológica

La sostenibilidad de este modelo depende de que las empresas de inteligencia artificial moneticen sus desarrollos antes de que el hardware quede obsoleto. Los procesadores actuales pierden valor comercial de forma acelerada cuando aparecen nuevas generaciones más eficientes.

Si la rentabilidad del sector se ralentiza y las tecnológicas deciden renegociar sus contratos de alquiler, la empresa pantalla dejaría de abonar los cuotas de la deuda. Esto afectaría directamente a los rendimientos de la aseguradora en Bermudas, que se vería obligada a liquidar activos de calidad para cubrir las pérdidas de unos componentes usados cuyo valor de reventa en el mercado secundario sería residual.

La estructura se mantiene dentro de la legalidad vigente y permite a cada participante obtener beneficios inmediatos: el fabricante de microchips incrementa sus ventas, la tecnológica accede a infraestructura sin endeudarse y la gestora cobra comisiones por intermediación. El riesgo final, sin embargo, permanece concentrado en los fondos de los asegurados que sostienen la base de la inversión.

https://x.com/USronaldcarter/status/2061159657570242723

Saludos.

[pollo]

Esto es lo que realmente es la IA señores. Un tinglado financiero creado por gentuza megalómana que no dudaría en montar una dictadura, sostenido por un producto que no cumple expectativas y es extremadamente ineficiente pero quienes mandan (tanto en empresas como en gobiernos) no tienen la capacidad o el conocimiento para percibirlo.

Cuando esto se caiga por su propio peso, habrá que ver qué ocurre en EE.UU.

https://www.euribor.com.es/2026/06/02/el-preocupante-entramado-financiero-que-sostiene-la-burbuja-de-la-inteligencia-artificial/

El preocupante entramado financiero que sostiene la burbuja de la inteligencia artificial

Carlos López · 2026.06.02

Cuando Michael Burry habla, Wall Street escucha con atención. El gestor de fondos que se hizo de oro por anticipar la crisis de las hipotecas basura en 2008 ha puesto ahora el foco en el sector tecnológico. Según su análisis, detrás de los ingresos récord de la inteligencia artificial se esconde un entramado de ingeniería financiera que traslada el riesgo de las tecnológicas hacia los fondos de pensiones de los ciudadanos.

Este mecanismo permite que empresas como Nvidia y la filial xAI de Elon Musk muevan miles de millones de euros en equipamiento tecnológico sin reflejar deudas en sus balances, utilizando una estructura legal que recuerda a los complejos productos que provocaron la crisis de 2008.

La creación de empresas pantalla para limpiar el balance

El proceso comienza con una venta de procesadores de última generación por valor de 5.400 millones de dólares. Estos chips son esenciales para entrenar modelos de lenguaje avanzados, pero en lugar de venderse directamente a una gran tecnológica, la operación se realiza a través de una Sociedad de Propósito Específico llamada Valor.

Valor es una empresa fantasma sin empleados ni actividad propia, diseñada exclusivamente para figurar como propietaria legal de los componentes. Los chips se envían directamente a los centros de datos de xAI, la empresa de Elon Musk, que los utiliza en régimen de alquiler mediante un contrato de arrendamiento neto triple. De este modo, la tecnológica asume los costes de mantenimiento y los impuestos, pero evita registrar la deuda de la compra en sus libros contables.

Por su parte, Nvidia registra la operación como una venta cerrada, lo que impulsa sus ingresos trimestrales ante los inversores. Sin embargo, el análisis revela que Nvidia financió parte de la operación aportando 1.900 millones de dólares a la propia empresa pantalla. En el sector, esta práctica se conoce como financiación circular: aportar fondos a tu propio comprador para inflar las cifras de ventas.

El papel del crédito privado en la sombra

Para cubrir los 3.500 millones de dólares restantes, la estructura recurre al mercado de crédito privado a través de Apollo Global Management. Tras las regulaciones bancarias impuestas tras la crisis anterior, las entidades tradicionales evitan financiar operaciones de alto riesgo respaldadas por activos que se deprecian rápidamente. Las firmas de capital riesgo han asumido ese papel, operando con una menor supervisión regulatoria.

Apollo emite títulos de deuda respaldados por los pagos del alquiler de los chips y el valor de los propios procesadores. Posteriormente, vende estos activos a Athene, una aseguradora de su propiedad especializada en la venta de rentas vitalicias para la jubilación.

Los fondos de jubilación como soporte del sector tecnológico

A través de esta conexión, los ahorros de los ciudadanos terminan financiando la adquisición de infraestructura tecnológica de alto riesgo. Los clientes que contratan estos productos buscan estabilidad y asumen que su dinero se invierte en activos seguros, como bonos del Estado. En su lugar, sus fondos sostienen la expansión de servidores informáticos de última generación.

Para evitar que los reguladores nacionales exijan reservas de capital elevadas por mantener activos ilíquidos en cartera, la aseguradora traslada la gestión de estos fondos a filiales radicadas en Bermudas, donde la normativa de solvencia es más flexible.

Activos sin precio de mercado y un apalancamiento elevado

En esta jurisdicción de baja regulación, una tercera parte de la cartera de la aseguradora se compone de activos de Nivel 3. Según las normas contables, estos productos carecen de un precio de mercado observable, por lo que su valoración se calcula mediante modelos matemáticos internos de la propia empresa, un método que en el sector se denomina de forma irónica valoración imaginaria.

El riesgo se incrementa debido a que la estructura opera con un apalancamiento de dieciséis veces su capital. Esto implica que una reducción del 6,25% en el valor real de los procesadores consumiría por completo los fondos propios que sostienen el entramado, comprometiendo la solvencia del sistema.

El impacto de la obsolescencia tecnológica

La sostenibilidad de este modelo depende de que las empresas de inteligencia artificial moneticen sus desarrollos antes de que el hardware quede obsoleto. Los procesadores actuales pierden valor comercial de forma acelerada cuando aparecen nuevas generaciones más eficientes.

Si la rentabilidad del sector se ralentiza y las tecnológicas deciden renegociar sus contratos de alquiler, la empresa pantalla dejaría de abonar los cuotas de la deuda. Esto afectaría directamente a los rendimientos de la aseguradora en Bermudas, que se vería obligada a liquidar activos de calidad para cubrir las pérdidas de unos componentes usados cuyo valor de reventa en el mercado secundario sería residual.

La estructura se mantiene dentro de la legalidad vigente y permite a cada participante obtener beneficios inmediatos: el fabricante de microchips incrementa sus ventas, la tecnológica accede a infraestructura sin endeudarse y la gestora cobra comisiones por intermediación. El riesgo final, sin embargo, permanece concentrado en los fondos de los asegurados que sostienen la base de la inversión.

https://x.com/USronaldcarter/status/2061159657570242723

Saludos.

[muyuu]

https://techcrunch.com/2026/06/05/google-will-pay-spacex-920m-per-month-for-compute/

Es curioso que esto lo haga SpaceX y no sea parte del negocio de xAI, pero supongo que hay que inflar la IPO.

Véase:

https://arstechnica.com/tech-policy/2026/06/sp-500-blocks-fast-spacex-entry-wont-waive-rule-for-unprofitable-ai-firms/

S&P 500 rejects SpaceX, also blocking entry for OpenAI and Anthropic
SpaceX won’t get easy access to billions of dollars from passive investors.

Jeremy Hsu · 2026.06.05

SpaceX signage outside the Space Exploration Technologies Corp. facility in Hawthorne, California, on June 3, 2026. Credit: Michael Yanow/NurPhoto via Getty Images

SpaceX has requested unusually swift entry into several leading stock market indexes as a condition of its historic stock market debut. But the S&P 500 stock market index representing many of the largest profitable US companies has surprised market analysts by refusing to bend the rules for Elon Musk’s space and AI company.

The June 4 decision by S&P Dow Jones Indices —the company that creates and manages stock market indexes such as the S&P 500— means that SpaceX will not gain accelerated access to potentially billions more dollars through passive investment funds that automatically purchase shares of S&P 500 companies. An exception for SpaceX could have also allowed leading AI companies such as OpenAI and Anthropic to gain entry not long after their own expected initial public offerings (IPOs). That possibility has now been shuttered.

The news will likely come as a relief to people concerned about passive investor money and people’s retirement savings plans having greater exposure to the market risks associated with SpaceX’s big bet on AI and speculative orbital data center plans. AI companies are generally facing more challenges in funding and building expensive AI data centers, even as they shift more of the subsidized costs of running AI services onto shocked customers through usage-based pricing.

To weigh expedited entry for SpaceX, the S&P Dow Jones Indices held a monthlong consultation to consider changing or waiving several main requirements for so-called MegaCap companies with “unprecedented market capitalizations”.

Those proposed changes included shortening the “seasoning period” for new IPOs from 12 months to six months, waiving the investable weight factor (IWF) requirement for MegaCap companies to make at least 10 percent of their shares publicly available, and waiving the requirements for MegaCap companies to demonstrate profitability in the latest quarter of the financial year along with the previous four quarters.

Such rule changes would have accommodated SpaceX’s plan to only offer approximately 3 percent of its IPO shares to public investors, and the fact that SpaceX is currently unprofitable with a growing debt load that has reached $29 billion because of its spending spree on AI infrastructure.

But in its final decision, the S&P Dow Jones Indices stated that “no changes will be made to the eligibility criteria including financial viability screens, seasoning period, or minimum IWF”. Even after the standard yearlong wait, SpaceX, Anthropic, and OpenAI may struggle to deliver the consistent profitability necessary to qualify for the S&P 500.

Money rules and exceptions

Swift entry into the S&P 500 would have triggered $14 billion of passive fund buying for SpaceX, according to Bloomberg Intelligence. The investment research arm of Bloomberg also estimated that OpenAI could have gained more than $8 billion, and Anthropic could have netted $4.6 billion from similar passive buying sprees triggered by their S&P 500 entries.

This is because $7.5 trillion in passively managed funds—popular among both individual investors and institutional investors—follow the S&P 500 by purchasing shares of companies according to their proportional representation in the S&P 500 index. For example, the Vanguard and Fidelity brokerage giants both offer passive investment funds that track the S&P 500 composition.

However, the S&P Dow Jones Indices did “carve out one concession” by changing the investable weight factor rules for “lower-profile benchmarks” such as the S&P Total Market Index and Dow Jones US Total Stock Market Index, according to Quartz. That could allow an IPO faster entry into those indexes.

By contrast, the Nasdaq stock exchange changed its rules to allow SpaceX to enter the Nasdaq-100 Index within 15 trading days as opposed to the usual three months. Similarly, the FTSE Russell index provider decided to give SpaceX and other follow-on companies  to the Russell Top 500 Index after the close of the fifth trading day following an IPO.

The denial of accelerated S&P 500 entry for SpaceX comes just days after Morningstar analysts described SpaceX as having been “significantly overvalued” in the lead-up to its IPO. The investment research firm valued SpaceX at $780 billion—less than half of SpaceX’s $1.75 trillion IPO goal—primarily based on the strengths of SpaceX’s Starlink satellite service and rocket launch business.

Saludos.

[el malo]

https://www.xataka.com/robotica-e-ia/problema-no-gastar-muchos-tokens-que-mayoria-se-estan-desperdiciando

Del "tokenmaxxing" hemos pasado al "tokenwasting": el nivel de malgasto en IA está alcanzando cotas inéditas

El problema no es gastar muchos tokens, es que la mayoría se están desperdiciando

Hace un año, Sam Altman hacía una predicción llamativa: a medida que la producción de centros de datos se automatice, el coste de la inteligencia [IA] debería converger en algún momento con el coste de la electricidad". O lo que es lo mismo: el acceso a la IA sería muy, muy barato. Eso no ha ocurrido ni mucho menos, pero es que además de gastar muchos tokens, los estamos desperdiciando.

Tanta IA para qué. El fenómeno del tokenmaxxing —el consumo desenfrenado de tokens más como moda que como algo útil— ha comenzado a disparar las alarmas, porque las empresas se han dado cuenta de que están gastando pequeñas fortunas para que sus empleados intenten sacar todo el partido posible a la IA.

Despedicio de IA. Un estudio de la startup EntelligenceAI afirma que por cada dólar invertido en IA, solo 18 centavos acaban llegando a producción. EL 82% restante se acaba invirtiendo en corregir errores, reescribir código y ejecutar procesos de revisión que no generan valor directo. Es lo que llaman "gasto improductivo", y es una señal de alerta porque el éxito de esta tecnología no depende de que usemos la IA sin parar, sino de que la usemos para mejorar la productividad.

Uber avisa. Andrew Macdonald, COO de Uber, cuestionaba abiertamente si ese gasto masivo de empresas como la suya en IA está realmente justificado cuando no está ligado a mejoras en la productividad. La empresa ha sido una de las que ha decidido cortar el gasto en los modelos de Anthropic porque ya se había "ventilado" el presupuesto anual disponible para usarlos.



La incertidumbre está ahí. Otros expertos advierten justo de lo contrario: esto es solo el principio de lo que está por venir, así que tomar medidas contra el consumo de IA puede ser contraproducente. El problema no está tanto en que se esté usando la IA, sino en que se esté desperdiciando: esa obsesión por consumir tokens hizo que en Amazon por ejemplo el CFO les dijese a sus empleados "No uséis la IA solo por el mero hecho de usarla". La empresa premiaba a aquellos que usaban más la IA, así que muchos acabaron usándola para tareas triviales, redundantes o inútiles.

Usar la IA de forma adecuada. Matan Gringberg, CEO de la startup de IA Factory, contaba en WSJ cómo un directivo de una importante institución financiera le había contado que sus empleados estaban gastando cientos de miles de dólares al mes en tokens. El problema estaba en que algunos estaban usando los modelos más potentes para contestar preguntas sencillas o simplemente para charlar. El mensaje aquí es claro: hay que usar estos modelos de forma adecuada para no malgastarlos: "Si tu hija necesita clases particulares de álgebra, probablemente puedas encontrar a alguien más barato que Albert Einstein para dárselas", concluía.

Estamos consumiendo tokens por encima de nuestras posibilidades. En el evento Google I/O Sundar Pichai, CEO de Alphabet, explicó que la empresa procesa en la actualidad más de 3.200 billones de tokens al mes, siete veces más que hace un año. Ante esa demanda, tanto ella como otras empresas están "castigando" el uso trivial de los modelos de IA.

Los agentes de IA consumen tokens como si no hubiera mañana. Lo que también ha ocurrido es que la llegada y popularización de las herramientas agénticas de programación, como Claude Code, Codex o Antigravity, hace que se consuman muchísimos más tokens porque con ellas es posible automatizar la ejecución de tareas de programación (o de otros ámbitos) de forma continua. El modelo de IA prepara un plan, lo ejecuta y a cada paso piensa y evalúa sus respuestas antes de continuar con el plan. Ese proceso es intensivo en consumo de tokens, y es el principal motivo de que los consumos de tokens se hayan disparado.

De tarifas planas, nada. Los planes mensuales como ChatGPT Plus o Claude Pro ofrecían margen de maniobra para los desarrolladores a la hora de consumir enormes cantidades de tokens sin apenas limitaciones. Sin embargo, tanto OpenAI como Anthropic y otras empresas han comenzado a cambiar sus estrategias, limitando los casos en los que se pueden usar esas tarifas plana para que los usuarios no puedan abusar de ellas. Si quieren consumir más pueden, pero siempre mediante una filosofía de pago por uso: cuanto más usas, más pagas algo que al menos ayuda a que los usuarios sean conscientes que no pueden usar modelos súperpotentes para conversaciones inútiles con sus chatbots.

[muyuu]

dos artículos largos:


https://www.wheresyoured.at/ai-is-slowing-down/

Last week I went on Bloomberg and discussed the state of the AI bubble with a clarity that rattled even the sweatiest boosters, mostly because I spoke with clarity about an investment frenzy whipped up through hype, deceit and mythology. Some were equal parts frustrated and angry that I don’t have money in the market, or, as they’d put it, “skin in the game.”

I get it! When your entire worldview is dictated by what a series of venture capitalists and psuedo-journalists on Twitter want you to believe, it must be difficult to imagine someone having “morals” or “beliefs” or that one might hold a position that wasn’t entirely based on greed or tribalism. It must be confusing — upsetting, even! — to hear that somebody is willing to accurately and vociferously tear into a tech industry largely controlled by people with no regard for their users or workers, who are willing to bathe their products in mediocrity all because it’s the thing that everybody else is doing.

This is a hysterical era perpetuated by liars, cowards, imbeciles, craven boosters and the easily-fooled. Those excited about generative AI are either the victim or the perpetrator of a con centered around a technology to ingratiate at the highest cost possible.

AI Cannot Afford To Slow Down — It Needs $3 Trillion Or More In Revenue By End Of 2030 To Sustain Its Existence
I also think that everybody is a little flippant about what has to happen for me to be wrong.

Whatever obtuse fantasies you have about the current state of generative AI are irrelevant to a much larger problem: that the infrastructure being built and compute commitments being made are being done so at a level that demands that generative AI and AI compute generate over $2 trillion in annual revenue by 2030. When I say that, I mean it absolutely has to do that otherwise none of the data center capex makes sense, and neither Anthropic nor OpenAI can pay their commitments.

OpenAI expects to spend $50 billion on compute in 2026, and I wouldn’t be surprised if Anthropic spends anywhere from $30 billion to $50 billion. Between them, Anthropic and OpenAI represent the vast majority of all AI compute demand — at a minimum 70%, if not 80% to 90%.

Put another way, there’s barely a few billion dollars of demand outside of two companies that lose billions — or tens of billions — of dollars a year.

Let’s break down these numbers a little further:

190GW of data center capacity assuming a PUE of 1.35 suggests a critical IT load of around 140GW, which, charged at around $12.5 million per megawatt, works out to around $1.75 trillion in annual revenue.

OpenAI and Anthropic project to make $184 billion and $174 billion in revenue in 2029, for a total of $358 billion in annual revenue. While Anthropic claims it will be profitable by then, I do not believe it will be, nor is it profitable at this point outside of financial engineering.

At present, there are no other major purchasers of AI compute outside of NVIDIA, hyperscalers (who are selling it to Anthropic and OpenAI, or they’re Meta, which has no AI strategy), OpenAI, and Anthropic. None. I can’t find a single one outside of Jane Street spending more than a few hundred million. We need a few hundred billion.

That’s already a huge problem, but the other problem is that we also need companies to spend dramatically more on AI services than they already do. While journalists are currently gooning over OpenAI and Anthropic making $6 billion or $10 billion in a given quarter, that’s just not enough! Both Anthropic and OpenAI need to be making $10 billion or more in monthly revenue by Q1 2028, or their growth rates aren’t going to support their compute commitments.

This is not hyperbole! Every single thing I have stated here precisely maps to the projections and promises of the AI industry. No matter how horny or flaccid you are for the potential of AI, it must grow at an astonishing, unstoppable rate from here until the end of time to be anything close to worthy of its costs.

Actually, sorry, let’s put judgments aside for a second, because this isn’t about judgment, but rather the promises that have been made by the software and hardware companies associated with AI. NVIDIA’s place atop the stock market and its ridiculous projections depend on both the continued flow of data center debt and the continued belief that AI services will have the revenue to back it up.

AI cannot, under any circumstances, slow down. In a year, Anthropic and OpenAI’s businesses have to be roughly twice the size they are today, and then double again basically every year until 2029 or 2030. In that time period, they must also both raise hundreds of billions of dollars or, alternatively, turn deeply unprofitable businesses into profitable ones while also doubling their revenues.

Alternatively, both must severely reduce their costs…except if they do that, they won’t have any need for all that compute capacity, which will deprive Oracle, Google, Microsoft, SpaceX, Cerebras, CoreWeave, TeraWulf, Cipher, and Hut8 of the $1.1 trillion in remaining performance obligations.

Also, if OpenAI can’t afford — or doesn’t want — its compute, Oracle will simply run out of money. It is spending anywhere from $340 billion to $700 billion (depending on whether you believe Jensen Huang in September 2025 or May 2026) on the 7.1GW of data centers it’s building for OpenAI. These, again, are not hyperbolic statements, but the actual costs associated with Oracle’s massive buildouts in Michigan, New Mexico, Wisconsin and Texas. I didn’t agree to do this! Larry Ellison did!

(continúa)


https://martinalderson.com/posts/xais-new-rental-business/

xAI is looking more like a datacentre REIT than a frontier lab


xAI is renting huge amounts of GPU capacity to Anthropic and Google. Financial engineering ahead of the SpaceX IPO, a real compute shortage, or a genuine datacentre advantage? Probably all three.



An unexpected development over the past few weeks is xAI's new partnerships with Anthropic and Google, providing them with a huge amount of capacity. It's worth remembering that xAI is now part of SpaceX, after the two merged back in February - so the revenue from these deals flows straight into the entity about to go public. While much has been made of the potential financial engineering given SpaceX's upcoming IPO, I think there's a bit more to this than just pure accounting tricks.

Anthropic was in a serious bind
If you use Claude products much, you'll be (very, probably) aware that Anthropic has had serious capacity problems, especially early afternoon onwards in Europe and in the mornings in the US (this is when demand seems to be highest as both European users and the Americas are both at work, fighting for capacity). I've written about this compute crunch before a few times - the coming crunch, whether it's here yet, and what comes next.

This resulted in Anthropic having to introduce new peak hour restrictions on their subscriptions, with usage between 5am–11am PT / 1pm–7pm GMT using more of your usage limit - with the aim of smoothing demand between peak hours and off peak hours where they had more capacity available.

However, there is only so much demand shifting you can do when demand is growing as fast as Anthropic's. At some point you end up having to ration users further, which definitely is far from ideal when you have both Google and OpenAI breathing down your neck for customers.

xAI to the rescue?
At the start of May, xAI announced a partnership with Anthropic to provide access to their (older) Colossus 1 datacentre in Memphis. This allowed Anthropic to reverse the usage limit restrictions on their subscriptions, and in general while stability of Anthropic services still leaves a lot to be desired, the peak time crunch has abated (for now, at least).

The fees involved are enormous, ramping to $1.25bn/month for 300MW of capacity - approximately 220k GPUs.

Last week, Google announced a similar partnership - $920mn/month for 110k GPUs[1]. It's important to note that both agreements have cancellation clauses - allowing either party to cancel with 90 days' notice after an initial lock-in period.

If you take this on face value, this is a ludicrously profitable deal for xAI

(continúa)