Bernstein analysts remain bullish on the Bitcoin mining sector, as deals with third-party providers will be necessary to address the computing power limits of AI data centers. The investment manager’s Bitcoin mining industry deal tracker registered a new AI-related deal every week in July, with combined deals standing at more than 7.5 gigawatts or the contracted equivalent of $150 billion in multi-year contracts, according to a Thursday research note.

The analysts said that third-party computing capacity from Bitcoin miners will remain highly valuable as access to power remains the AI industry’s real bottleneck amid growing political pushback against building new US data centers. Bitcoin mining stocks logged double-digit gains on Monday after Bitcoin mining companies Hut 8 and IREN announced major AI infrastructure deals. Hut 8 announced a 15-year, $9.8 billion lease for its AI data center campus and IREN disclosed $2.8 billion in cloud services contracts with AI developers.

“IREN is beginning to convert that infrastructure advantage into contracted and more predictable revenue,” said Seeking Alpha contributor The Curious Analyst on Thursday. “The biggest risk to my thesis is execution,” they said. The contributor rates IREN a strong buy.

July sees slew of AI-miner tie-ups

Other publicly traded Bitcoin mining companies have also expanded into AI. Earlier in July, MARA Holdings announced plans to acquire a Texas site with up to 2 gigawatts of capacity to expand its AI and digital infrastructure business. Days earlier, TeraWulf signed a 20-year data center lease with AI startup Anthropic, a deal the company said could generate roughly $19 billion in contract revenue. Bitcoin mining infrastructure company Bitdeer has also expanded into AI cloud services and high-performance computing.

Most of the miner stocks were poised for gains on Thursday, based on premarket activity. HUT 8’s shares were up 5.23%, IREN was up 1.89% and TeraWulf was up 1.49%. Sector tracking exchange-traded fund CoinShares Bitcoin Mining ETF (WGMI) was up 1.47% ahead of the Nasdaq open. Bernstein has an outperform rating on all of the stocks it named, except for MARA, which it rates as market perform.

The surge in AI-mining partnerships is not limited to these larger players. Smaller mining firms are also exploring similar arrangements, realizing that their existing infrastructure—built for energy-intensive Bitcoin mining—can be repurposed for AI workloads. This has created a new revenue stream for miners at a time when Bitcoin mining profitability has been under pressure due to halving events and rising network difficulty.

AI data centers face political pushback

Bernstein’s research note said that Bitcoin miners and third-party computing power providers will remain important for AI companies, as the construction of new data centers is facing growing bipartisan political pushback. On Wednesday, Texas Democratic Senate candidate James Talarico reportedly shared a proposal to create stronger local approval processes and to repeal tax breaks for AI data centers. In April, US Senator Ron Wyden shared concerns that AI data centers in his home state of Oregon could worsen water scarcity during persistent droughts. He said that large data centers can consume up to 5 million gallons of water per day and asked large data center operators to explain how they would reduce groundwater withdrawals to protect the local water supply.

In March, President Donald Trump’s administration published a Ratepayer Protection Pledge aiming to expand AI infrastructure without increasing electricity bills for households and small businesses. In January, several state governors published plans to expand the grid to meet the rapidly growing demand from AI data centers, but stated that new data centers should bear the costs they create, rather than shifting these to existing residential customers and small businesses.

The political headwinds underscore the critical role Bitcoin miners can play. Miners have already secured large blocks of power capacity in regions with cheap electricity, often in rural areas or industrial zones where local opposition is less intense. They also have experience managing variable power loads and can quickly adjust operations to sell power back to the grid when needed—a flexibility that traditional data centers lack. This makes existing mining sites attractive for AI companies that need to deploy computing capacity rapidly without waiting years for new power infrastructure.

Moreover, the environmental concerns surrounding AI data centers are not dissimilar from those that have long plagued Bitcoin mining. Critics have pointed to the water consumption and carbon footprint of large-scale computing. However, as Bernstein notes, repurposing existing mining infrastructure can be more efficient than building from scratch. Many miners are also transitioning to renewable energy sources, further aligning with the sustainability goals of AI firms.

Technical and economic drivers

Behind the headlines lies a fundamental shift in the economics of computing. The explosion of generative AI models has created an insatiable demand for graphics processing units (GPUs) and other specialized chips. Traditional data centers operated by cloud giants like Amazon Web Services and Microsoft Azure are being supplemented by new, purpose-built facilities. But the lead time for constructing a major data center can be three to five years, while the demand for AI compute is immediate. Bitcoin miners, who already operate thousands of high-powered computing rigs in purpose-built facilities, can refit their sites to host AI servers relatively quickly.

This convergence has been accelerated by the maturation of the Bitcoin mining industry, which has consolidated into large public companies with sophisticated financial operations. These firms have strong balance sheets and access to capital markets, allowing them to fund the infrastructure upgrades needed for AI workloads. The deals tracked by Bernstein, including the 7.5 GW of capacity under contract, represent a fraction of what is possible. Analysts estimate that Bitcoin miners globally control over 20 GW of power capacity, much of which could be converted for AI use.

The financial implications are significant. For miners, AI contracts provide long-term, predictable revenue streams that are uncorrelated with Bitcoin price volatility. For AI developers, they offer a way to bypass the bottleneck of data center construction and secure the computing power needed to train and run large models. Bernstein’s overweight rating on the sector reflects this symbiotic relationship, which they believe will continue to deepen as the AI industry grows.

However, risks remain. Execution is key, as The Curious Analyst noted. Converting a Bitcoin mining facility to an AI data center requires substantial capital expenditure and technical expertise, including cooling systems, network infrastructure, and security. There is also competition from traditional data center developers who are increasingly looking to build on former industrial sites. Yet Bernstein’s data suggests that the market believes the miners have a first-mover advantage that will persist.

In addition to the specific deals mentioned, the broader ecosystem is evolving. Bitdeer, for example, is not only offering cloud services but also developing its own chips for AI workloads. MARA Holdings’ acquisition in Texas is intended to create a combined Bitcoin mining and AI campus that can flex between the two uses. This hybrid model could become the industry norm, allowing miners to maximize utilization of their power assets regardless of market cycles in either Bitcoin or AI.

The political landscape will continue to shape the opportunity. While states like Texas are generally pro-business, local concerns about grid reliability and water usage are mounting. The proposal by James Talarico in Texas and Senator Wyden’s scrutiny in Oregon are signs that even before the federal ratepayer pledge, states are taking action. Bitcoin miners, by positioning themselves as part of the solution rather than part of the problem, may be able to navigate these regulatory challenges. The flexibility to curtail operations during peak grid stress is a strong selling point.

Looking ahead, Bernstein expects the deal flow to accelerate into 2027 and beyond. As more AI models require ever-larger training clusters, the demand for ready-to-use compute will only increase. Bitcoin miners that have already made the pivot to AI are likely to see upgraded valuations, similar to how cloud computing transformed the fortunes of traditional data center REITs a decade ago. For now, the market is rewarding the early movers, and the analysts see further upside as the narrative becomes mainstream.

Source:Cointelegraph News