2026 Bitcoin Mining Winter: CleanSpark, MARA, Riot Earnings Breakdown of the Loss Reality and AI Transformation Divergence Path

As Bitcoin’s price hovers around $79,337, significantly below the approximately $126,000 peak set in October 2025, Bitcoin mining company stocks—once considered the “most leveraged bull market”—are facing a harsh earnings season test. CleanSpark’s net loss widened nearly twofold year-over-year to $378.3 million, MARA’s quarterly loss reached $1.26 billion, while Riot took the lead in realizing data center revenue—these three leading miners’ earnings reports are not only a test of their operational capabilities but also reflect the structural divergence across the entire Bitcoin mining industry amid deepening halving effects and hashprice approaching historic lows.

An Industry Winter Behind a Loss Report

In early May 2026, CleanSpark (Nasdaq: CLSK), MARA Holdings (Nasdaq: MARA), and Riot Platforms (Nasdaq: RIOT) successively disclosed their first-quarter 2026 financial results. All three reported net losses of varying degrees, drawing widespread market attention.

CleanSpark reported a quarterly net loss of $378.3 million, with a basic loss per share of $1.52, compared to a net loss of $138.8 million in the same period last year. MARA’s quarterly net loss was about $1.26 billion, versus $533.4 million a year earlier. Riot also was not spared, with a quarterly net loss of $500.5 million and a diluted loss per share of $1.44, far exceeding analyst expectations of a $0.72 loss.

The release dates of these three financial reports are highly concentrated, and combined with Bitcoin’s continued low prices in Q1 2026, this earnings season quickly became a focal point of industry sentiment.

From Halving Bonuses to Industry-Wide Contraction

To understand the deeper implications of these earnings reports, we need to revisit Bitcoin’s fourth halving event in April 2024. The halving reduced block rewards from 6.25 BTC to 3.125 BTC, directly halving miners’ revenue per block produced. In the roughly one year following the halving, Bitcoin’s price surged—rising from about $63,000 to a cycle high of around $126,000 in October 2025—creating a lucrative profit window for miners.

However, the situation sharply reversed after Q4 2025. Bitcoin’s price fell from about $124,500 (early October 2025) to roughly $86,000 (end of December 2025), a decline of about 31%. Meanwhile, total network hash rate continued to climb, and hashprice—the metric measuring daily income per unit of hashpower—kept decreasing.

Key facts at critical points in this cycle include:

• April 2024: Bitcoin completes its fourth halving, reducing block rewards to 3.125 BTC
• October 2025: Bitcoin hits a cycle high of about $126,000
• Q4 2025: Weighted average cash mining cost rises to about $80,000, hashprice drops to approximately $36–38 per PH/s/day
• February 2026: Hashprice falls to about $28 per PH/s/day, hitting a new low post-halving
• March 20, 2026: Bitcoin mining difficulty decreases by about 7.7%, one of the largest drops of the year
• May 14, 2026: Bitcoin price is $79,337.4, a 22.08% change over the past year (based on Gate market data)

This timeline clearly shows that miners enjoyed the profit bonanza from Bitcoin’s price rise after halving in the first half of 2025, but since the second half of 2025, falling prices combined with rising hash rate have sharply deteriorated profitability. The massive losses in Q1 2026 are essentially a financial reflection of this trend.

Data and Structural Analysis: Who Is “Naked Swimming”?

A cross-comparison of the financial data of the three miners reveals significant differences in their situations.

Key Financial Data Comparison of the Three Major Miners Q1 2026

All data sourced from official financial disclosures (up to March 31, 2026 quarter) and cross-verified from multiple sources:

Data sources: Company filings and cross-verification

CleanSpark: Strategic Bet Amid Revenue Contraction

CleanSpark’s quarterly revenue was $136.4 million, down 24.9% YoY. Of its net loss of $378.3 million, about $224.1 million stems from fair value changes in Bitcoin holdings. As of March 31, the company held $260.3 million in cash, with Bitcoin holdings valued at $925.2 million, total assets of $2.9 billion, and long-term debt around $1.8 billion.

CleanSpark’s financials show typical “book losses diverging from operational reality.” Excluding fair value changes, its operational pressures mainly stem from: first, revenue shrinking due to lower Bitcoin prices; second, a significant increase in depreciation and amortization costs, reflecting upfront costs from fleet expansion.

In a persistently low hashprice environment, CleanSpark’s debt expansion strategy faces a dilemma—if its AI infrastructure leasing business cannot generate short-term revenue, high leverage could turn into liquidity pressure.

MARA: Active Balance Sheet Reshaping

MARA’s quarterly revenue was $174.6 million, down 18.3% YoY. About $100 million of its net loss results from unrealized fair value adjustments on Bitcoin holdings (non-cash). During the quarter, MARA sold 20,880 BTC, cashing out about $1.5 billion at an average price of roughly $70,137, with about $1 billion used to repurchase convertible notes, reducing outstanding convertible debt by about 30%. By quarter-end, MARA held 35,303 BTC, remaining the fourth-largest corporate Bitcoin holder.

According to US FASB ASU 2023-08, from fiscal year 2025 onward, companies must measure crypto assets at fair value, with Bitcoin price fluctuations directly impacting earnings. The huge losses reported by these miners are related to this accounting standard.

While MARA’s losses are large, they are essentially “paper losses” driven by accounting rules, not cash flow. Its sale of Bitcoin and repurchase of convertible debt are effectively measures to reduce future dilution risk.

Although financially rational, MARA’s large Bitcoin sales might be interpreted by the market as a lack of confidence in short-term Bitcoin prices, adding downward pressure on its stock.

Riot: First to Realize Data Center Revenue

Riot is the only one among the three with YoY revenue growth, at 3.6%, reaching $167.2 million. Data center revenue contributed $32.2 million—comprising $0.9 million from operational leasing and $31.2 million from tenant fit-out services. Engineering revenue was $22 million. Mining revenue was $111.9 million, down 21.7% YoY. AMD expanded its contracted capacity with Riot from 25 MW to 50 MW during the quarter.

Though data center revenue accounts for only about 20% of total revenue, its significance lies in testing whether “miners transforming into AI data center operators” can generate actual income. AMD’s expansion indicates initial market recognition of this transformation. After the earnings release, Riot’s stock rose about 10%, contrasting with declines in the other two miners.

If Riot can continue expanding data center contracts and generate stable leasing income over the next few quarters, its valuation logic might shift from “cyclical miner” (high beta) to “infrastructure operator” (steady cash flow, predictable).

Market Opinions: Why Such Divergence?

Market views around this earnings season are sharply divided, centered on three key issues.

Dispute 1: Are Book Losses Over-Interpreted?

Mainstream View 1 (“Noise Theory”): The huge losses of the three miners mainly come from fair value adjustments of Bitcoin holdings, which are non-cash items and should not be used to assess operational quality. For example, MARA’s roughly $1 billion loss is essentially a product of accounting standards (FASB fair value rules), unrelated to cash flow.

Mainstream View 2 (“Signal Theory”): Even excluding fair value changes, all three companies’ core mining revenues have declined, reflecting the fragility of a business model relying solely on Bitcoin mining amid persistent hashprice declines.

While CleanSpark and MARA’s fair value losses are non-cash, the fact remains that CleanSpark’s revenue fell 24.9% YoY and MARA’s revenue fell 18.3% YoY—facts that cannot be ignored. There is a real tension between book profits and operational reality.

Dispute 2: Is AI Transition a Path or Just a Narrative?

The power infrastructure, cooling systems, and site resources owned by miners are precisely the scarce assets needed for AI data centers. According to CoinShares, the market assigns higher valuation multiples to miners with AI infrastructure concepts—up to 12.3x, far above pure miners. The report projects that by the end of 2026, AI-related business could contribute up to 70% of revenue.

AI data centers differ significantly from Bitcoin mining farms. They demand higher standards for power stability, network latency, and cooling, with substantial retrofit costs. Not all mining sites are suitable for AI deployment.

Currently, Riot is the only one generating substantial data center revenue ($33.2 million), while MARA’s AI deployment is still in development—its Long Ridge acquisition plan is expected to see initial AI capacity online by mid-2028. CleanSpark’s AI/HPC infrastructure is also in early stages. The narrative of AI transformation carries a premium in capital markets, but actual revenue realization still requires time.

Dispute 3: Are Miner Stocks Undervalued?

Since 2026, miner stocks have rebounded significantly. Hut 8 and Riot, for example, are up about 85% and 46% respectively this year. Despite Bitcoin’s relatively flat performance, the market’s expectations for AI-driven transformation are decoupling stocks from Bitcoin prices.

Whether this rally is supported by fundamentals remains doubtful. After MARA’s earnings, its stock fell about 5%, closing at $12.65, and dropped another 1.85% after hours. CleanSpark’s stock also faced pressure post-earnings. This indicates that market focus on financial data persists despite the AI narrative.

While year-to-date, miner stocks outperform Bitcoin, this “excess return” is more driven by AI-driven valuation re-pricing rather than real fundamental improvement. If AI progress falls short of expectations, valuation corrections could be substantial.

Industry Impact Analysis: Accelerating Miner Segmentation

This earnings season sends a clear signal: the business model divergence among Bitcoin miners is accelerating, with profound implications for industry structure.

The Winner-Loser Battle in Hashrate

CoinShares’ Q1 2026 report shows that about 15–20% of global Bitcoin miners are operating at a loss at hashprice levels of roughly $28–33 per PH/s/day, mainly those with outdated hardware or high electricity costs. On March 20, 2026, Bitcoin difficulty dropped about 7.7%, one of the largest reductions of the year, indicating some miners have already exited the network. As of May 2026, difficulty has been lowered three times in a row (a first since July 2022).

This “miner surrender” phenomenon, while posing some short-term security risks to the Bitcoin network, from an industry cleanup perspective, helps remove inefficient capacity, restoring supply-demand balance and creating healthier profit margins for remaining miners.

Capital Side: Debt Play

The three miners’ capital structure choices differ markedly. MARA has sold Bitcoin to reduce convertible debt, lowering future dilution risk. CleanSpark has taken the opposite approach, issuing zero-coupon convertible preferred notes to expand financing, with long-term debt reaching about $1.8 billion. Riot has adopted a more balanced strategy, selling some Bitcoin and entering credit agreements to maintain liquidity.

These differing capital strategies reflect their respective outlooks: MARA’s deleveraging indicates cautious management regarding near-term conditions, while CleanSpark’s leverage expansion bets on infrastructure assets delivering outsized future returns.

Revenue Structure Divergence

The most fundamental divergence lies in revenue composition. Riot has begun generating substantial income from AI data centers, MARA is advancing its transformation via acquisitions like Long Ridge, while CleanSpark maintains its core Bitcoin mining focus but is also cautiously exploring AI/HPC opportunities.

The next 12–18 months will be critical to validate the AI transformation narrative. Companies capable of generating predictable, sustainable income from AI infrastructure will see their valuation frameworks shift from “cyclical miner” (high beta) to “infrastructure operator” (steady cash flow, predictability); pure miners lagging behind will continue to bear the high beta risks associated with hashprice volatility and Bitcoin price cycles.

Is Bitcoin Mining Still Worth Investing?

Based on the above data analysis and industry trends, here are multiple scenario evolutions regarding whether “Bitcoin mining is still worth investing in.”

Scenario 1: Moderate Bitcoin Price Recovery (assume BTC rises to $90,000–$100,000)

In this scenario, hashprice could rebound from about $33 to $40–$45 per PH/s/day, enabling most mid-to-high-end hardware miners to regain profitability. Miners with diversified AI income streams (like Riot) will benefit from both “improved mining margins + AI revenue growth.” Pure miners (like CleanSpark) will also see improved profits, but their stock price elasticity may be weaker than peers with AI premium.

CoinShares notes that hashprice is highly sensitive to BTC price, and with less efficient miners having exited earlier, remaining miners’ unit hashpower income will be further boosted.

Scenario 2: Continued Bitcoin Weakness (assume BTC stays below $80,000 long-term)

If Bitcoin remains below $80,000, CoinShares forecasts hashprice could decline further, forcing more miners to exit. In this case, highly leveraged miners like CleanSpark will face significant debt pressures; MARA, having already deleveraged substantially, is more resilient; Riot’s data center revenue can serve as a “cushion.” Pure miners may face a “double whammy” of valuation and profit declines.

Currently, about 15–20% of total hashpower is unprofitable; if hashprice drops further, exit scale could accelerate.

Scenario 3: Large-Scale AI Transition Realized (assume 2027, AI revenue exceeds 50%)

If miners’ AI data center businesses proceed as planned and generate substantial recurring income, the entire sector’s valuation logic will be reshaped. Investing in miners would shift from betting solely on Bitcoin prices to allocating to digital infrastructure. Leading AI-capable miners (Riot, MARA) could achieve valuation multiples similar to data center REITs, rather than cyclical miner multiples.

Over $70 billion in AI/HPC contracts have been announced, indicating strong demand for AI infrastructure. However, execution risk and time lag between contract announcement and revenue realization remain.

Scenario 4: Bitcoin Enters a New Bull Market

If Bitcoin embarks on a new rally, the fair value of miners’ Bitcoin holdings will turn positive, and financials will improve significantly. But this depends on whether the market still views miners as “Bitcoin proxy investment tools”—in an environment where AI narratives strengthen, the correlation between miner stocks and Bitcoin prices may weaken.

Conclusion

As of May 14, 2026, with Bitcoin at $79,337.4 (based on Gate market data), the earnings reports of CleanSpark, MARA, and Riot present a highly fragmented industry landscape. Behind the net losses are factors such as business model choices, capital strategies, and industry cycle positioning.

The question “Is Bitcoin mining still worth investing in?” cannot be answered simply with “yes” or “no” in 2026. A more precise question is: Do investors seek high beta exposure to Bitcoin, or do they prefer a digital asset company actively transitioning into AI infrastructure?

If the former, miner stocks face even greater uncertainty in the context of halving effects and hashprice pressures, with their “bullish leverage” attributes under threat more than ever. If the latter, each company’s AI progress, data center contract pipeline, and capital structure must be evaluated individually, because the label “miner” no longer fully captures the complexity of this group.

Three halvings, one bull cycle, one deep correction—Bitcoin mining is at a pivotal point from “extensive expansion” to “focused differentiation.” The ultimate winners will not be determined by reported losses but by the real execution of transformation over the next 12 to 18 months.