Could MARA’s Bitcoin mining sites become a shortcut to the AI data center boom that is reshaping Big Tech?

MARA Holdings is betting that the infrastructure built for Bitcoin mining could become valuable for a very different reason: artificial intelligence. Its strategy is increasingly focused on power, land, and data-center infrastructure that could help AI companies bring computing capacity online faster.

The idea is not as simple as turning off Bitcoin miners and plugging in GPUs. MARA is trying to use years of investment in energized sites, power access, and digital infrastructure as a starting point for AI and high-performance computing, while Bitcoin mining remains part of the business.

Could Bitcoin mining sites really give MARA an AI shortcut?

Yes, but the shortcut is mainly about infrastructure rather than computing hardware. MARA has spent years securing power and developing large computing sites, giving it assets that can be difficult for new AI data-center developers to assemble from scratch.

That matters because electricity has become one of the biggest constraints on new AI infrastructure. The U.S. Energy Information Administration expects American electricity demand to reach record levels in 2026 and 2027, with data centers and AI among the forces driving that growth.

MARA said its energized mining capacity reached 70.3 EH/s in the second quarter of 2026, while its broader power portfolio could reach about 4.8 gigawatts after pending transactions. The company argues that controlling power-ready sites gives it a foundation for the next generation of computing infrastructure.

The distinction between power-ready and AI-ready is important, however. A site with electricity, land, and an existing mining operation still needs specialized cooling, networking, electrical systems, computing equipment, and other infrastructure before it can support demanding AI workloads.

That makes MARA’s strategy less about replacing one type of computer with another and more about reusing an existing infrastructure platform. The company can potentially avoid some of the early challenges involved in finding land, securing power, and building a connection to the grid.

The broader market is also placing greater value on those assets because AI developers are competing for access to electricity. Reuters recently reported that the next industrial race is increasingly centered on the ability to produce and distribute electricity as digital infrastructure expands.

Fun fact: The deal marks a sharp reversal in the policy of a company that has spent the past six months aggressively selling assets to fund a large-scale expansion into artificial intelligence.

Why does power matter more than ever for AI?

Power is becoming a central issue because AI data centers can require enormous amounts of electricity and take years to develop. MARA itself described power as the defining constraint for AI deployments in its second-quarter shareholder letter.

That changes the value proposition for Bitcoin mining sites. Bitcoin miners traditionally searched for inexpensive electricity because power is one of their highest operating costs, but those same power arrangements can become attractive to companies looking for large computing campuses.

MARA’s Long Ridge transaction illustrates this strategy particularly well. The company agreed to acquire a campus in Hannibal, Ohio, with a 505 MW combined-cycle gas plant, more than 1,600 acres, and access to power, water, fiber, and grid infrastructure.

The site is expected to provide more than 1 GW of total potential power capacity and has a path toward as much as 600 gross MW of AI and critical IT loads. MARA said initial AI and critical IT construction is expected to begin in the first half of 2027, with initial capacity targeted for service around mid-2028.

Those details show why the real asset may not be the Bitcoin mining machines themselves. It is the collection of infrastructure surrounding them, including land, power generation, interconnection and existing digital facilities.

That advantage is becoming more important as developers face grid delays and increasing competition for electricity. Reuters reported that some U.S. states are now questioning huge data-center power requests because many proposed projects may not have the funding or technical capacity needed to reach completion.

Is a gigawatt announcement the same as an AI data center?

No, and that distinction could become one of the most important details for investors and readers to watch. MARA’s announced power portfolio represents potential infrastructure capacity, not several gigawatts of completed AI computing capacity operating today.

The company’s Starwood partnership is designed to help move selected sites through pre-development and eventually into data-center development. The agreement includes pre-development services, leasing work, and development structures, showing that the transition requires several steps before a site becomes an operating AI facility.

MARA’s June 2026 filing also shows that the Starwood projects were still in the pre-development process. The company had approved up to $136.3 million in pursuit costs across the projects, which highlights the amount of work that can occur before a facility reaches construction.

That makes customer contracts another important milestone. MARA said it was working with Starwood to secure at least two leases before the end of 2026, meaning signed customers could provide a clearer indication of progress than a headline power target alone.

Readers should therefore separate several stages of the AI buildout. A site can have potential power capacity, energized power, planned IT capacity, contracted capacity and operating AI capacity, and those figures do not mean the same thing.

This also explains why MARA’s AI transition could take years even though the company already owns or controls valuable infrastructure. The shortcut is potentially skipping part of the site-acquisition and power-development process, not eliminating the engineering and construction needed to operate an AI data center.

Fun fact: MARA is working with Starwood to reduce construction risk and expects to sign two leases by year-end while evaluating both large campuses and smaller inference-focused sites.

How could MARA keep Bitcoin mining while building AI capacity?

MARA does not necessarily have to abandon Bitcoin mining while developing AI infrastructure. Its strategy allows flexible computing workloads, meaning Bitcoin mining can remain part of the portfolio while higher-value AI and HPC projects are developed.

This approach gives the company a way to keep generating revenue from existing infrastructure during a long construction cycle. Instead of waiting for an entire facility to become an AI campus, portions of a site can potentially continue operating as mining capacity while other areas are prepared for new workloads.

MARA has also continued to treat Bitcoin as an important financial asset. At the end of June, the company held 35,577 BTC, including Bitcoin that was loaned or pledged as collateral, with the holdings valued at about $2.1 billion at that time.

The company has used Bitcoin as a source of financial flexibility while pursuing its infrastructure strategy. MARA said it had begun selling Bitcoin to fund operations and expected to continue monetizing holdings when market conditions and capital-allocation priorities made that useful.

The latest reported purchase of 1,292 BTC for roughly $98.6 million therefore adds an interesting twist to the story. It suggests that MARA’s strategy should not necessarily be viewed as a one-way exit from Bitcoin, especially when the company continues to operate a large mining business.

Instead, MARA appears to be pursuing a hybrid infrastructure model. Bitcoin can remain one workload and an important balance-sheet asset while AI and HPC create another potential source of revenue from the same broader infrastructure platform.

What makes Long Ridge different from a typical Bitcoin mine?

Long Ridge shows how far MARA’s strategy can extend beyond simply converting a mining building. The Ohio campus combines power generation, land, water, fuel supply, fiber, grid interconnection, and existing digital infrastructure in one location.

That combination could reduce some of the development risks associated with starting an AI campus from an empty piece of land. MARA specifically said assets with this combination of power, land, and infrastructure are becoming harder to replicate because securing permits, interconnection, and power can take significant time.

The company is also expanding beyond Ohio. MARA said it secured rights to a roughly 2 GW power-ready land site in Matagorda County, Texas, subject to ERCOT and interconnection approvals, which could help expand its overall power portfolio.

Starwood adds another piece to the strategy because it brings data-center development and leasing expertise to MARA’s infrastructure. MARA said Starwood has experience delivering infrastructure for hyperscalers and frontier AI companies, while the partnership is intended to provide development expertise and proportional capital support.

MARA is also developing capabilities beyond physical sites through Exaion, which the company describes as part of its effort to participate across multiple layers of the AI infrastructure value chain. That suggests the company wants to build a broader digital infrastructure business rather than depend entirely on Bitcoin mining or one type of data-center customer.

The bigger question is whether these assets can turn into contracted and operating AI capacity. Until leases are signed, construction advances, and customers begin using the infrastructure, much of the AI opportunity remains a development pipeline rather than completed revenue-producing capacity.

Could MARA’s power strategy reshape its business?

MARA’s strategy could change the way the market views a Bitcoin miner because the company’s infrastructure may have value beyond cryptocurrency mining. The key asset could increasingly become access to power and the ability to turn that power into different types of computing capacity.

That shift reflects a broader change in the economics of digital infrastructure. As AI companies compete for electricity, sites that already have power access, land, and grid connections can become strategically important even when their original purpose was Bitcoin mining.

MARA still faces significant execution challenges, including construction costs, financing, permitting, customer acquisition, and the technical requirements of AI workloads. The company’s own filings make clear that development projects involve substantial costs and that future results depend on completing transactions and turning planned infrastructure into operating capacity.

The company’s recent Bitcoin activity also shows why the transition is unlikely to be completely linear. MARA can continue using Bitcoin mining, its Bitcoin holdings, and infrastructure financing to support its broader strategy while waiting for AI projects to reach commercial operation.

For Big Tech and AI infrastructure developers, the attraction is easier to understand when the problem is viewed from the power side. A Bitcoin miner that already controls a large amount of energized infrastructure may offer a faster starting point than a developer that still needs to find land, secure electricity, and navigate grid connections.

MARA’s biggest test will therefore be execution rather than the size of its announced power portfolio. If it can turn existing mining infrastructure into contracted AI capacity, the assets built for Bitcoin could become an important bridge into the data-center economy that is rapidly emerging around artificial intelligence.

TL;DR

• MARA is using its Bitcoin mining infrastructure, including power, land, and grid access, as a potential shortcut into the growing AI data-center market.
• The strategy does not mean abandoning Bitcoin mining, as MARA can continue mining while developing AI and high-performance computing capacity.
• Its Long Ridge campus in Ohio shows how power generation, land, water, fiber and existing infrastructure could support future AI workloads.
• MARA’s large power portfolio is valuable, but announced gigawatts should not be confused with completed or operational AI data-center capacity.
• Starwood is helping MARA with development and customer leasing, while projects still face construction, financing, permitting, and customer-acquisition challenges.
• The key test will be whether MARA can turn its existing infrastructure into signed AI leases and operating capacity rather than simply announcing large future power targets.

This article was made with AI assistance and human editing.

If you liked this, you might also like:

• Why Mexican authorities just raided a secretive crypto mine hidden in the mountains
• Why Revolut mistakenly handed over customer Bitcoin data after a fake official request
• Why Anthropic just chose Nasdaq for what could become one of the biggest IPOs ever