Anthropic turns to Gulf states for frontier AI cash

Anthropic has long positioned itself as the ethical, safety-first darling of Silicon Valley. But when the price tag of frontier AI reaches into the hundreds of billions, even the purest principles face a brutal reality check.

According to a leaked Slack memo obtained by WIRED, Anthropic CEO Dario Amodei told staff that the company is preparing to seek investments from the United Arab Emirates and Qatar. Addressing the elephant in the room, Amodei acknowledged that taking money from Middle Eastern leaders would likely enrich dictators. He admitted he was not thrilled about it, adding that believing no bad person should ever benefit from our success is a pretty difficult principle to run a business on.

Chasing a $100B capital pool

The driving force behind the pivot is simple: staying competitive at the cutting edge requires mind-boggling amounts of cash. Amodei noted that the Middle East holds easily $100 billion or more in available capital. In his view, Anthropic gains an enormous advantage by accessing that pool, and without it, remaining on the frontier becomes substantially harder.

That marks a stark contrast to the company's past hesitation. In 2024, Anthropic reportedly turned away Saudi capital over national security concerns when FTX's stake was auctioned off, though UAE-backed ATIC ultimately bought a major chunk. And in May, when Donald Trump led a tech delegation including Sam Altman, Elon Musk, and Jensen Huang to Saudi Arabia and the UAE, Anthropic sat it out.

Meanwhile, rivals have moved aggressively. OpenAI partnered with UAE-based MGX on its massive $500 billion Stargate infrastructure initiative and revealed plans for an Abu Dhabi data center. Watching competitors lock up gigawatt-scale funding clearly forced Anthropic's hand.

Money without the hardware

To justify the move internally, Amodei insisted on strict guardrails. Anthropic is pursuing what he called a narrowly scoped, purely financial investment to prevent foreign backers from gaining operational leverage.

He reaffirmed his fierce opposition to building massive training clusters in the Middle East or shipping Nvidia H20 chips to China, arguing that handing the AI supply chain to authoritarian regimes poses severe national security risks. Amodei did acknowledge the risk of soft power, admitting that the implicit promise of future funding might make it harder to push back later. His proposed game plan: take as many billions as possible without compromising, and hold the line if investors demand more.

Amodei also expressed frustration with the broader industry. He noted that Anthropic pushed hard against allowing giant data centers in the region, but without collective rules, the race to the bottom took over. Competitors now seem fine outsourcing massive 5-gigawatt training runs to the Gulf, leaving Anthropic at a severe disadvantage. He even vented about the inevitable comms backlash, calling the media and Twitter quick to scream hypocrisy while misunderstanding the substantive stakes.

Why it matters

For developers and enterprise users relying on Claude, this move does not change model safety guidelines or day-to-day access. But it sends a crystal-clear message: developing frontier models has grown far too expensive for traditional venture capital and American tech giants alone. If Anthropic wants Claude to compete with GPT-5 and beyond, it has to tap sovereign wealth.

Across the industry, the Gulf states have officially cemented their status as the kingmakers of frontier AI. The competition is no longer just about algorithms or talent; it is about who can marshal gigawatts of power and sovereign-level bankrolls.

In our view, Amodei's memo is remarkable for its cynical transparency. Setting up a firewall between passive financial checks and actual compute infrastructure is a clever compromise that lets Anthropic keep its chip supply chain on home soil. Yet soft power is notoriously hard to quarantine. Once you open the door to sovereign money, holding the line on future demands will test Anthropic's safety identity like never before.