Research from Bernstein released today indicates that Core Scientific’s reported 75% return on its high-performance compute deal with CoreWeave is not a model for the broader bitcoin mining sector. The firm pegs expected stabilized returns for peers TeraWulf and Cipher Mining at just 5% and 4%, respectively, noting Core Scientific’s figure is an outlier due to its established infrastructure. The analysis arrives alongside a positive move in Bitcoin’s price, which traded at $65,156 as of 14:05 UTC today, a 2.34% gain on 24-hour volume of $31.45 billion.
Context — why this matters now
The push for bitcoin miners to diversify into high-performance compute for artificial intelligence workloads became a dominant narrative in 2025. This transition followed the 2024 Bitcoin halving, which reduced block rewards and pressured miners to find new revenue streams. Several public miners announced partnerships with AI firms to repurpose their large-scale data centers and secure power contracts for more profitable compute tasks.
This pivot is happening against a backdrop of rising energy costs and persistent volatility in Bitcoin’s price, which currently holds a market capitalization of $1.31 trillion. The immediate catalyst for evaluating these AI deals is the recent wave of earnings reports and new partnership announcements, which have provided concrete financial details beyond initial press releases. Investors are now scrutinizing the actual return profiles of these announced ventures.
Data — what the numbers show
Bernstein’s analysis provides concrete figures that reveal a stark divergence in projected profitability. While Core Scientific’s deal with CoreWeave is forecast to generate a 75% return on assets, Bernstein models a stabilized return on assets of just 5% for TeraWulf and 4% for Cipher Mining. This indicates a 70 to 71 percentage point gulf between the top and median performers in the sector.
A comparison of key metrics highlights the disparity. Core Scientific’s advantage stems from owning fully depreciated mining infrastructure and possessing power contracts at fixed, low rates, allowing it to capture most of the revenue from AI compute. TeraWulf and Cipher, while also deploying capital toward AI, face higher marginal costs for power and infrastructure build-out, compressing their net returns. The sector's performance contrasts with Bitcoin's own market moves, which saw a 24-hour gain of 2.34%.
Analysis — what it means for markets / sectors / tickers
The analysis suggests a clear bifurcation in the mining equity sector. Tickers like CORZ may see sustained premium valuations if its high-margin AI revenue materializes as projected, while WULF and CIFR could face pressure as the market discounts their more modest return profiles. Companies with stranded power assets or older, depreciated facilities are better positioned to benefit from the AI compute shift than newer miners with higher debt loads.
A key counter-argument is that AI demand itself could rise sufficiently to lift all boats, improving pricing power for all providers of accelerated compute. The primary risk is execution; building and operating AI data centers requires different technical expertise than Bitcoin mining, and delays or cost overruns could erase thin margins. Recent market flow shows investors rotating capital toward miners with proven AI revenue streams and away from those with only aspirational announcements.
Outlook — what to watch next
The next major catalyst for the sector is Core Scientific’s Q2 2026 earnings report, expected in early August, which will provide the first detailed financials inclusive of its CoreWeave deal. Investors will also watch for updates from TeraWulf and Cipher Mining on their AI capacity ramp-up timelines during their respective earnings calls. Key levels to monitor include Bitcoin’s support near $63,000 and its year-to-date high around $67,500, as crypto market sentiment heavily influences miner equity liquidity.
If AI revenue for TeraWulf and Cipher exceeds the conservative 4-5% ROA projections, a re-rating of those stocks is likely. Conversely, failure to meet these targets could lead to significant underperformance. The broader macro environment, including Federal Reserve policy decisions on interest rates, will also impact the cost of capital for future infrastructure expansions across the industry.
Frequently Asked Questions
What is a stabilized return on assets (ROA) in this context?
Stabilized ROA refers to the projected annual return a company earns on its deployed assets after the initial build-out phase is complete and operations are running at normal capacity. For bitcoin miners entering AI, this metric factors in the capital expenditure for servers, ongoing energy costs, maintenance, and the revenue from selling compute cycles. A stabilized figure smooths out one-time costs to show the underlying business profitability.
How does Core Scientific’s infrastructure give it such a large advantage?
Core Scientific’s key advantage is its ownership of fully depreciated mining infrastructure from previous cycles. This means the buildings, cooling systems, and electrical work are already paid for and on its books at near-zero value. Combined with long-term, fixed-rate power contracts secured years ago, the company can allocate almost all new AI server revenue to profit, unlike peers who must cover high depreciation and current-market energy costs.
Are other bitcoin miners besides TeraWulf and Cipher pursuing AI deals?
Yes, several other public miners have announced AI or high-performance compute initiatives, including BitDigital, Hut 8, and Iris Energy. Each company’s potential returns will depend on its specific power costs, location, existing infrastructure, and the terms of its client contracts. The Bernstein report focuses on the most advanced public deals, but the return spectrum across the entire sector is likely just as wide, underscoring the need for deal-by-deal analysis.
Bottom Line
Bernstein's research underscores that AI deals are not a uniform value creator for bitcoin miners, with return profiles dictated by existing infrastructure and power costs.
Disclaimer: This article is for informational purposes only and does not constitute investment advice. CFD trading carries high risk of capital loss.
