Author: David, Deep Tide TechFlow
On March 1, Iranian missiles and drones struck the Gulf region, with one landing on an Amazon data center in the UAE.
The data center caught fire, lost power, and approximately 60 cloud services were interrupted.
Claude, one of the world's most widely used AIs, runs on Amazon's cloud. On the same day, Claude experienced a global outage.
Anthropic's official statement was that a surge in users overwhelmed the servers.
As of the time of writing, complaints about Claude's service being unavailable are still circulating on social media; the well-known prediction market Polymarket has already launched a prediction topic: "How many more times will Claude be down in March?".
If it is ultimately confirmed that Iran was responsible, this would be the first time in human history:
A commercial data center was physically destroyed in war.
But why would a civilian data center be bombed?
Rewind two days. On February 28, the US and Israel jointly launched an airstrike on Iran, killing Supreme Leader Khamenei and a number of senior officials.
A significant portion of the intelligence analysis, target identification, and battlefield simulation for this airstrike was done with the help of Claude. Through cooperation between the military and data analytics company Palantir, Claude had long been integrated into the US military's intelligence system.
Ironically, just hours before the airstrike, Trump ordered a comprehensive ban on Anthropic because Anthropic refused to hand over its AI to the Pentagon without restrictions. But despite the ban, the war had to be fought.
Publicly, it was said that it would take at least six months to remove Claude from the military system.
So before the ink on the ban was dry, the US military took Claude to bomb Iran. Then Iran retaliated, and a missile landed on the data center running Claude AI.
Image source: Bloomberg
The data center was most likely not targeted, merely caught in the crossfire. But regardless of whether the missile was aimed at the data center or not, one thing is certain:
Truth is within the range of the cannon, and AI is within the range of artillery. This applies to both the side firing the artillery and the side being shelled.
The Great AI Infrastructure, Built on the Middle East Powder Keg
Over the past three years, Silicon Valley has moved half of the AI industry to the Middle East Gulf.
The reason is simple. The UAE and Saudi Arabia have the world's wealthiest sovereign wealth funds, cheap electricity, and one regulation:
If you want to serve my customers, the data must be stored in my territory.
So Amazon opened data centers in the UAE and Bahrain each, and invested $5.3 billion in Saudi Arabia to build another; Microsoft has nodes in the UAE and Qatar, and its Saudi facility is also ready.
OpenAI, in collaboration with Nvidia and SoftBank, is building a $30 billion+ AI park in the UAE, touted as the largest computing base outside the US mainland.
In January this year, the US just signed an agreement with the UAE and Qatar called "Pax Silica". Translated, it means "Silicon Peace", which sounds beautiful.
The core content of the agreement is to control the flow of chips, ensuring that advanced chips do not fall into Chinese hands.
In exchange, the UAE obtained a license to import hundreds of thousands of Nvidia's most advanced processors annually. Abu Dhabi's G42 cut ties with Huawei, Saudi AI companies promised not to buy Huawei equipment...
The entire Gulf's AI infrastructure, from chips to data centers to models, has comprehensively leaned towards the US.
These agreements considered everything, from chip export controls, data sovereignty, investment reciprocity, to technology leakage risks.
But none considered that someone would use a missile to bomb a data center.
An international security scholar at Qatar University said something quite fitting after seeing the Amazon data center fire:
"These security frameworks were designed for supply chain control and political alignment; physical security was never on the agenda."
Cloud computing has been telling a story for ten years about elasticity, redundancy, and decentralization. But data centers are buildings with addresses, with walls, roofs, and coordinates. No matter how advanced your chips are, if the data center is bombed, it's bombed.
"Cloud" is a metaphor; data centers are not.
AI seems virtual, running in code, floating in the cloud. But code runs on chips, chips are installed in data centers, and data centers are built on Earth.
Who Protects AI?
This Amazon data center can be said to have been affected, or optimistically, collateral damage.
But what about next time?
In the context of escalating global geopolitical conflicts, if your data center is running AI models that help an opponent with target identification, the opponent has every reason to treat your data center as a military facility to strike.
International law has no answer to this question either.
Existing laws of war have provisions for "dual-use facilities," but those clauses were written for factories and bridges; no one thought about data centers.
Is a data center that helps banks process transactions during the day and runs intelligence analysis for the military at night considered civilian or military?
In peacetime, data center选址 considers latency, electricity prices, policy incentives... When war comes, none of this matters. What matters is how far your data center is from the nearest military base.
So, this bombing has started to shift everyone's attention.
Previously, everyone was discussing the same anxiety: will AI replace my job? But no one discussed another question:
Before AI replaces you, how vulnerable is it itself?
A regional conflict paralyzed the Middle East node of the world's largest cloud service provider for a full day; and this was just one data center.
There are now nearly 1,300 hyperscale data centers worldwide, with another 770 under construction. These centers are consuming more and more electricity, water, and money, and承载着越来越多的事物——your deposits, your medical records, your food delivery orders, even a country's military intelligence...
But the方案 for protecting these data centers, to this day, is probably still fire suppression systems and backup generators.
When AI becomes a country's infrastructure, its security is no longer the responsibility of a single company. Who protects AI? Cloud providers? The US Pentagon? Or the UAE's air defense system?
This question was theoretical three days ago. Not anymore.
AI is within the range of artillery. Actually, it's not just AI. In this era, what isn't within the range of artillery?
