Anthropic запускает масштабный проект по созданию ИИ-инфраструктуры в США за $50 млрд Translation: Anthropic launches a massive project to create AI infrastructure in the U.S. for $50 billion.

The startup Anthropic has unveiled an ambitious $50 billion initiative aimed at developing AI infrastructure in the United States. The project will commence with the establishment of specialized data centers in Texas and New York.

This initiative is being carried out in collaboration with Fluidstack, a cloud platform for artificial intelligence that provides large-scale graphics processing clusters to various clients, including Meta, Midjourney, and Mistral.

The first facilities are expected to be operational by 2026, with the project anticipated to generate 800 permanent jobs and an additional 2,400 positions in the construction sector.

«We are on the verge of realizing artificial intelligence that can accelerate scientific discoveries and tackle complex problems in new ways. To unlock this potential, we need an infrastructure that can support ongoing development at the forefront,» stated Anthropic CEO Dario Amodei.

He emphasized that the new facilities will contribute to the creation of more powerful AI systems while also generating employment opportunities.

Anthropic’s announcement comes as competitors push forward with their own initiatives.

OpenAI has secured deals with Nvidia, Broadcom, Oracle, and other industry giants, totaling over $1 trillion for building facilities that will support AI technologies.

In turn, Alphabet, Meta, Microsoft, and Amazon have raised their capital expenditure forecasts to $380 billion by 2025. The scale of such investments is increasingly sparking concerns about a potential bubble in the sector.

Baidu founder Robin Li has pointed out that, so far, the primary beneficiaries of the AI boom are chip manufacturers rather than application developers.

Significant investments in AI infrastructure by startups have yet to yield returns. Anthropic currently serves over 300 enterprises, with the number of clients generating over $100,000 in revenue increasing nearly sevenfold over the past year. However, the company expects to reach break-even only by 2028.

Potentially, it could significantly outpace OpenAI, which is projecting operational losses of $74 billion in the same year due to rising computing costs.

The creator of ChatGPT anticipates that it will spend roughly 14 times more than Anthropic before achieving profitability by 2030.

High expenses aren’t hindering the growth of AI startups’ valuations, which are reaching trillion-dollar milestones. OpenAI is preparing for a public offering with an estimated valuation of $1 trillion, potentially positioning itself for one of the largest IPOs in history.

Large investments in infrastructure enable American AI startups to «play the long game» by establishing a stable and gradually increasing revenue stream. Chinese competitors, however, take a different approach, resulting in divergent outcomes.

As of August, only four AI applications from private companies in China have made it into the top 100 in terms of annual revenue. Glority, Plaud, ByteDance, and Zuoyebang collectively earned $447 million, which accounts for just 1.23% of the total revenue of $36.4 billion.

Rui Ma, co-founder of Tech Buzz China, explained the imbalance by citing the structure of China’s economy—local startups aim to generate quick profits, prioritizing short-term projects, such as those in the public sector (government contracts, municipal orders).

«Attracting global consumers and enterprises requires significantly more capital,» she noted.

Ma added that Chinese companies face challenges in raising funds in domestic markets.

Microsoft has chosen a different strategy and opted to leverage OpenAI’s efforts instead of making multi-billion dollar investments to strengthen its position in semiconductors.

The concept is as follows: Sam Altman’s startup is developing AI chips in partnership with Broadcom, while the creator of Windows utilizes the finished technologies.

«As they innovate even at the systems level, we gain access to everything,» explained Microsoft CEO Satya Nadella.

In October, Microsoft and OpenAI announced a revision of their partnership terms. Under the updated agreement, Microsoft’s stake in Altman’s AI startup is 27%, valued at $135 billion.

The revised contract solidifies Microsoft’s status as «a partner in advanced models» for OpenAI and maintains exclusive licenses for the intellectual property of the startup’s neural networks and products until 2032.

It is worth noting that in September, Anthropic introduced its new AI model, Claude Sonnet 4.5, which showcases advanced capabilities in programming. It is capable of creating «production-ready» applications.