Anthropic is making another massive move to secure the computing power it needs to keep up with demand.
The Claude maker has reportedly signed a $9.1 billion, 20-year deal with Riot Platforms for AI data center capacity at Riot’s Rockdale, Texas campus. The agreement gives Anthropic access to 191 megawatts of computing capacity through June 2048.
For Riot, this is more than just another customer. It is a major step in its transformation from a Bitcoin mining company into an AI infrastructure provider.
And for Anthropic, the deal highlights a bigger issue facing the AI industry: building powerful AI models is only half the battle. Having enough computing capacity to run them at scale is becoming just as important.
Anthropic Is Buying Computing Before It Runs Out
AI demand has been growing quickly, and Anthropic has been trying to secure enough infrastructure to keep its services running smoothly as more customers use Claude.
That has pushed the company into a series of huge infrastructure agreements.
Some of its recent commitments include:
- $9.1 billion with Riot Platforms
- $10 billion with Volta Infra Holdings
- Nearly $45 billion of computing capacity from xAI
The pattern is clear. Anthropic is not waiting for demand to slow down before securing infrastructure.
It is locking in computing capacity years ahead.
That matters because AI companies need enormous amounts of electricity, data center space, networking equipment and advanced chips to train and operate their models.
The challenge is no longer simply building a better AI model.
The challenge is having enough infrastructure to serve everyone who wants to use it.
Riot Is Turning Its Bitcoin Infrastructure Into an AI Business
The deal is particularly interesting because of who Anthropic is buying capacity from.
Riot is best known as a Bitcoin mining company. But crypto miners have increasingly found themselves sitting on something that is becoming extremely valuable in the AI economy: large-scale power infrastructure and data center sites.
Riot’s Rockdale, Texas campus already has significant power capacity and infrastructure.
Now, instead of using all of that capacity for Bitcoin mining, Riot is expanding into AI-focused computing.
The Anthropic agreement gives Riot a long-term customer and a predictable source of revenue.
The company said the contract is expected to generate $9.1 billion in revenue over 20 years.
There are also two options to extend the agreement by five years each, which could push the total value of the contract to as much as $16.1 billion.
That is a major change in the economics of Riot’s business.
191 Megawatts Is a Big Deal
The agreement covers 191 megawatts of computing capacity.
To put that into perspective, Riot said the amount of power involved is roughly equivalent to what 143,000 homes could consume at any given moment.
That gives you an idea of the scale of infrastructure required to support modern AI workloads.
AI data centers are extremely power-intensive. As models become larger and companies serve more users, the demand for computing capacity keeps rising.
This is why electricity availability, land, cooling systems and grid connections are becoming increasingly important parts of the AI investment story.
The AI race is increasingly becoming an infrastructure race.
Why Riot’s Transformation Matters
Riot’s move reflects a much broader shift across the crypto mining industry.
Bitcoin miners spent years building data centers in locations where they could access large amounts of relatively inexpensive electricity.
Now, AI companies are searching for many of those same advantages.
That creates an opportunity for miners to repurpose their infrastructure for AI workloads.
Riot is already pursuing this strategy.
The company has announced another agreement involving Advanced Micro Devices, as it builds out its data center business.
The strategy makes sense on paper.
Bitcoin mining can be highly dependent on cryptocurrency prices and energy costs. Long-term AI infrastructure contracts can potentially provide a more predictable revenue stream.
That could make AI data centers an increasingly attractive business for companies that already have the necessary power infrastructure.
Anthropic Isn’t the Only AI Company Chasing Capacity
Anthropic’s deal also needs to be viewed in the context of the wider AI infrastructure boom.
AI companies are competing for access to:
- Advanced chips
- Data center capacity
- Electricity
- Cooling systems
- High-speed networking
- Power generation and grid connections
The demand is so strong that infrastructure providers are becoming strategic partners for AI developers.
Anthropic’s recent deals show just how aggressively the company is preparing for future demand.
The company has already committed billions to secure computing capacity, rather than relying on infrastructure becoming available at the last minute.
That approach could become increasingly common as AI adoption expands.
The Bigger Investment Story: AI Is Spreading Beyond Semiconductors
For investors, one of the most interesting parts of this story is what it says about the AI supply chain.
The AI boom initially focused heavily on chipmakers.
Companies such as Nvidia and AMD became central to the story because advanced GPUs are essential for AI workloads.
But the infrastructure needed to use those chips is becoming just as important.
That includes the companies providing:
Power → Data centers → Cooling → Networking → Chips → Cloud infrastructure
Riot sits in the data center and power infrastructure part of that chain.
That means companies that were once considered part of entirely different industries could increasingly benefit from AI investment.
Crypto miners are one example.
Utilities, power developers, data center operators, networking companies and real estate owners with access to electricity could all become part of the AI infrastructure ecosystem.
Riot’s Stock Reaction Shows How Investors See the Shift
Investors responded strongly to the announcement.
Riot shares jumped 25% to $24.40 in late trading after news of the Anthropic agreement.
The reaction makes sense.
A 20-year agreement worth billions gives investors a clearer picture of how Riot’s infrastructure can potentially generate revenue beyond Bitcoin mining.
The company’s second-quarter sales also benefited partly from its data center business, another sign that the strategy is already becoming meaningful to the company.
But investors will still need to watch how quickly Riot can expand its AI infrastructure and how profitable that business ultimately becomes.
A large contract is important.
Turning that contract into reliable, profitable capacity is the next test.
What This Means for Anthropic
For Anthropic, the deal is primarily about securing enough computing power to support its growth.
Claude has become a major AI platform, and more users means more computing requirements.
The company cannot simply assume that data center capacity will always be available when needed.
By signing long-term agreements, Anthropic gets greater visibility into its future infrastructure.
There is also a strategic advantage.
If demand for AI continues to rise, having computing capacity secured years in advance could help Anthropic scale without constantly searching for new infrastructure.
But it also means making enormous long-term financial commitments.
Anthropic is effectively betting that demand for AI computing will remain strong for years.
The Bigger Question: How Much Power Will AI Need?
This deal brings the conversation back to one of the biggest questions surrounding the AI boom.
Where will all the electricity come from?
The industry is building AI models at an extraordinary pace, but those models ultimately need physical infrastructure.
That means more data centers.
More data centers mean more electricity.
And more electricity means more pressure on power grids, generation capacity and transmission infrastructure.
This is why the AI story is increasingly connected to the energy story.
The companies that can secure power and bring large data centers online may become just as important as the companies building the chips inside them.
What Investors Should Watch Next
The Riot-Anthropic deal raises several important questions for the market.
1. Can Riot successfully scale its AI data center business?
The company is moving beyond its traditional Bitcoin mining operations. Investors will want to see whether AI infrastructure can become a meaningful and profitable part of the business.
2. Will more Bitcoin miners follow the same path?
Riot is not alone in exploring AI data centers. If the economics continue to look attractive, other miners could begin shifting their infrastructure toward AI workloads.
3. Will AI companies continue signing multi-billion-dollar infrastructure deals?
Anthropic’s recent commitments suggest that major AI developers are willing to spend heavily to guarantee future computing capacity.
4. How much will power availability affect AI growth?
Access to chips is only one constraint. Electricity, cooling and data center construction could increasingly determine how quickly AI companies can expand.
5. Could AI infrastructure become a bigger investment theme than AI software?
The companies providing the physical foundation for AI may have a long runway if demand continues to grow.
The Bottom Line
Anthropic’s $9.1 billion deal with Riot Platforms is about much more than one AI company renting space from one Bitcoin miner.
It is another sign that the AI boom is changing the economics of the infrastructure around it.
For Anthropic, the priority is simple: secure enough computing power to meet demand.
For Riot, the opportunity is different: turn existing power and data center infrastructure into a new AI-focused revenue stream.
And for investors, the bigger story may be the shift happening underneath the AI boom.
The next phase of AI may not be determined only by who builds the smartest model.
It could also be determined by who has the power, data centers and computing capacity to run those models at scale.
The AI race is moving from the cloud to the power grid. And companies like Riot are positioning themselves right in the middle of that transition.