Anthropic invests $15 billion in Australia for "land grabbing" 1.4GW computing power; AI infrastructure arms race escapes US domestic bottleneck
```
As data center construction in the United States becomes mired in power shortages and community opposition, AI giants are turning their eyes overseas in search of new havens for computing power. Anthropic’s newly revealed overseas infrastructure plans mark the official cross-border escalation of the compute arms race among large model companies.
According to a Sunday report by the Australian Financial Review, Anthropic plans to secure at least 1.4GW of data center computing resources in Australia, with a projected total investment cost of up to $15 billion. According to a confidentially released tender document, the company aims to officially put at least 1GW of this computing power into use by the end of next year.
This capital expenditure, which far exceeds market expectations, is not only designed to support the exponential growth in its large model training and inference needs, but also reflects the industry pain point of the U.S. struggling to expand data centers at home. At a critical juncture of surging revenue and an impending IPO, Anthropic is building its ‘moat’ by securing physical infrastructure globally.
$15 billion "thrown" into the Southern Hemisphere, 1GW of compute delivered next year
Anthropic’s computing power layout in Australia demonstrates strong execution and a clear timeline. A scale of 1.4GW is equivalent to the output of several large nuclear power units, enough to support the full-capacity operation of hundreds of thousands of top-tier AI accelerator cards. The $15 billion construction cost also makes it one of the largest single AI infrastructure investments globally in recent years.
The tender documents show the project has adopted an aggressive delivery timetable, requiring at least 1GW of compute to be online by the end of next year. To support this strategy, Anthropic has recently been intensively recruiting data center engineers, electrical engineers, and project procurement specialists in Australia and Japan, with its infrastructure team accelerating expansion into the Asia-Pacific region.
Escaping domestic "NIMBY", Australia becomes a new oasis of computing power
Behind Anthropic’s land grab in the Southern Hemisphere lies the increasingly severe bottlenecks facing U.S. data center construction. Just recently, Blackstone-owned QTS announced it was terminating what was once touted as the “world’s largest” 850-hectare "Digital Gateway" project in Virginia, due to sustained local opposition and legal challenges.
As AI compute demand surges, traditional U.S. data center hubs face the triple threats of soaring electricity demand, water consumption, and community NIMBY ("Not In My Backyard") effects. In contrast, Australia—with its relatively abundant land, stable energy supplies, and suitable climate—stands out as an ideal alternative location.
Moreover, in March this year, Anthropic signed a memorandum of understanding (MOU) with the Australian government, agreeing to cooperate on AI safety research and the national AI agenda, paving a policy path for its large-scale infrastructure rollout.
Giant infrastructure arms race: from cloud leasing to physical land grab
Aggressive infrastructure expansion is a necessary move for Anthropic’s commercial breakout and acceleration in the capital markets. Currently, Anthropic's annualized revenue has soared from $9 billion at the start of the year to over $44 billion, with inference infrastructure gross margins rising above 70%, marking its first profitable quarter and rapid margin expansion.
Supported by strong financial performance, the company last month completed a $65 billion financing round, with a valuation of up to $965 billion, and has officially submitted IPO filings. However, iteration of large models and the proliferation of Agent applications are consuming massive amounts of compute. To prevent compute shortages from stifling its growth, Anthropic must lock in long-term, low-cost infrastructure ahead of time.
In the compute arms race, leading AI companies are shifting their strategies from simple cloud rental to deep involvement in physical infrastructure construction. Like rivals OpenAI and Google, Anthropic is building a diversified compute supply chain.
In addition to building physical data centers in Australia, Anthropic recently signed a nearly $45 billion cooperation agreement with SpaceX, reached an $1.8 billion agreement with Akamai, and continues to source chips and cloud services from Google. This “physical self-build + alliance with industry giants + cloud vendor leasing” hybrid model is not only a hedge against single-source supply chain risks, but also the fundamental base for AI giants to maintain their competitive advantage into the next technology cycle.
Risk Warning and DisclaimerThe market has risks, and investment requires caution. This article does not constitute personal investment advice and does not take into account any individual user’s special investment targets, financial situation, or needs. Users should consider whether any opinions, views, or conclusions in this article are suitable for their specific situation. Investments made on this basis are at your own risk. ```