Riot Platforms: A Value-Investing Deep Dive

Even so, the industry's profit pool is concentrated not in brand or distribution but in low power prices, machine efficiency, downtime management, sources of capital, and timing of monetization. Peers such as MARA, CleanSpark, IREN, and Cipher are likewise priced by the market on strong narratives and high volatility: as of May 20, 2026, MARA's market cap was about 4.73 billion dollars, CleanSpark's about 3.35 billion dollars, IREN's about 15.93 billion dollars, and Cipher's about 7.62 billion dollars; conventional P/E metrics for this group are mostly either negative or so high as to be uninformative. Put differently, this is not a contest of "good companies in a good industry," but rather a contest of "in a poor industry, who can survive the next cycle in better shape and tell a stronger new story." Opinion: industry attractiveness 2/5.

Moat and Management

Whether There Is a Moat at All

Broken down by Buffett's familiar moat categories, Riot's situation is not complicated.

The evidence behind these judgments is clear. Riot's cost advantage was once striking: in 2023, the company disclosed a cash cost to mine per BTC (excluding miner depreciation) of just 3,831 dollars, largely thanks to 71.215 million dollars of power-curtailment / demand-response credits. But by 2024 that figure had risen to 32,216 dollars, and in 2025 it rose further to 49,645 dollars. In other words, the so-called "low-cost moat" looks more like a cyclical advantage than a structural one that can be locked in for the long term. More pointedly, by Q1 2026 the all-in cost per BTC (including miner depreciation) had reached 96,283 dollars, while the production value per BTC that quarter was only 75,964 dollars. Without higher BTC prices or better hashrate / power efficiency, the current economics are uncomfortable.

What I think comes closest to a moat for Riot is not "mining" but the combination of power and land infrastructure it is gradually assembling in Texas. In January 2025, the company halted the originally planned 600 MW second phase of mining expansion at Corsicana, formally pivoting to evaluate AI/HPC uses, and expected to cut 2025 Corsicana capital expenditure by 245 million dollars as a result. It then, in February 2025, brought on new directors with experience in AI/HPC conversion, data centers, and real estate, and engaged Evercore, Northland, and others to court potential partners; and by January 2026 it had landed its first long-term lease with AMD. This sequence shows that management has recognized that the genuinely wider potential moat is not mining more BTC, but turning "power, land, interconnection, cooling, and construction capability" into assets it can lease to top-tier compute customers.

The catch lies exactly here: this moat is still under construction, not yet built. Competitors with enough financing capacity and suitable land and power can replicate it; AI/HPC customers ultimately want reliable power, cooling density, networking, delivery speed, and total cost, and will not pay a premium because of Riot's "brand." My conclusion is therefore: Riot's moat strength is 2/5, with the trend moving from a "weak traditional-mining moat" toward a "moderate infrastructure moat," but the validation is not yet complete.

Whether Management Can Be Trusted and Capital Allocation Is Rational

First the positive. In 2025, management halted the 600 MW second-phase mining expansion at Corsicana in favor of evaluating AI/HPC, and announced a corresponding cut to future capital expenditure; it then quickly signed the 25 MW AMD contract in early 2026, and disclosed in the Q1 2026 report that AMD had exercised a further 25 MW option, raising total contracted capacity to 50 MW. This at least shows management is not rigidly wedded to a single "pure mining" path, but is willing to shift toward more capital-efficient directions. With independent, constructive involvement from Starboard and D. E. Shaw, the board added three new directors, which also shows governance is not closed off.

Now the negative. Riot's capital-allocation record is far from pretty. First, persistent shareholder dilution is very clear: the company raised net proceeds of roughly 571.6 million dollars through ATM issuance in 2023, roughly 114.9 million dollars in 2024, and roughly 207.7 million dollars again through ATM in 2025; shares outstanding rose from 230.8 million at year-end 2023 to 344.9 million at year-end 2024 and 371.6 million at year-end 2025, reaching 379.0 million by March 31, 2026. From year-end 2023 to Q1 2026 alone, the share count expanded by roughly 64%. Second, the Bitfarms investment / contest did not end well: in 2024, Riot spent roughly 203.8 million dollars to buy about 90.10 million shares of Bitfarms, and in 2025 sold the entire stake for only about 106.1 million dollars in recovered proceeds, a clearly value-destructive capital allocation.

Now incentives and governance. The 2025 proxy statement shows Jason Les beneficially owned 8.375 million shares, Benjamin Yi beneficially owned 9.218 million shares, and all directors and officers together beneficially owned 24.284 million shares, about 6.93% of the share count at the time. This shows management is not a set of "zero-ownership professional managers," and alignment is not absent, but it is also far from the very strong binding of "founders with a high ownership stake." At the same time, the proxy disclosed 2024 CEO total compensation of roughly 83.53 million dollars; although it is mostly equity awards tied to performance milestones, that magnitude will still keep long-term shareholders watchful about dilution and incentive design.

On "honesty and accounting controls," I give a below-average grade. In fiscal 2022, the company had a material weakness related to controls around Bitcoin impairment review; but by 2024 and 2025, both management and Deloitte concluded that internal controls were effective. My reading is that there is not enough evidence to support a heavy charge like "fraud," but neither has this been a consistently "completely reassuring, mature financial culture." On balance, management and capital allocation score 2/5.

Financial Quality and Cash Flow

The conclusion first: Riot's revenue is growing fast, but its financial quality is not high. Revenue growth has not translated smoothly into distributable cash flow; instead it has come alongside heavy capital expenditure, miner prepayments, equity dilution, and, more recently, added debt. For a value investor, this is not a "the bigger it grows the easier it gets" financial trajectory, but more of a "the bigger it grows the more it needs continuous investment and a continuous next-chapter story" trajectory.

The table below uses the four complete fiscal years 2022–2025 plus Q1 2026. I deliberately do not stretch further back, because Riot's current business shape only took form in recent years, and earlier years are of limited comparability to today.

Note: revenue, net income, operating cash flow, capital expenditure on property and equipment, share count, and equity all come from the company's original annual and quarterly disclosures; "Reported FCF" and "Economic FCF" are calculated from these, with "Economic FCF" treating miner prepayments as capital expenditure as well, since that spending is essentially payment for future capacity.

Looking only at revenue, Riot is tempting: 2022–2025 revenue CAGR was about 35.7%. But the picture is entirely different on net income and cash flow. The 2024 profit does not mean the business model has matured, because the profit was mixed with Bitcoin fair-value gains, an industry tailwind, and power-strategy credits; then in 2025 it swung to a massive loss of 663.2 million dollars, and in Q1 2026 lost a further 500.5 million dollars. For a high-quality business, such a profit trajectory leans too heavily on accounting and cyclicality rather than on a stable underlying base.

Now the cash. Operating cash flow was -255.1 million dollars in 2024, -572.9 million dollars in 2025, and a further -182.7 million dollars in Q1 2026. Even on the "Reported FCF" basis, the company's free cash flow has been persistently and substantially negative in recent years; if miner prepayments are included as "economic capital expenditure," the cash burn is heavier still. This shows Riot's growth is not "compounding growth driven by internal cash flow," but "expansion supported by external financing, asset sales, or BTC-inventory monetization."

The balance sheet is not in immediate danger, but it is not comfortable either. As of March 31, 2026, the company had cash and restricted cash of roughly 282.5 million dollars, a combined carrying value of Bitcoin and restricted Bitcoin of roughly 1,069.7 million dollars, total debt of roughly 842.2 million dollars, total liabilities of roughly 1,043.3 million dollars, and shareholders' equity of roughly 2,394.7 million dollars. If BTC is treated as a quasi-cash asset, the company still has positive "net financial assets"; but if you regard BTC as operating inventory / a high-volatility asset, then the company has in fact moved from being "nearly unlevered" a few years ago to one that now needs BTC collateral and credit lines to support expansion. In particular, the 200 million dollar Coinbase credit line had been fully drawn as of Q1 2026, at an interest rate of about 8.3%, collateralized by 5,802 BTC.

On earnings quality, two facts deserve special attention. First, 2025 EBITDA was -306.4 million dollars, while 2024 EBITDA was 297.0 million dollars, showing extreme profit elasticity; second, 2025 interest expense was 24.14 million dollars, up sharply from 1.985 million dollars in 2024, indicating that the cost of leverage is starting to enter the statements. For a high-quality enterprise, interest should be a rounding item; for Riot, it is no longer a footnote that can be ignored.

Overall judgment: profit does not equal real cash profit; growth clearly requires large capital investment; the company is not "easier as it grows" but more like "needing more money as it grows"; there is no sufficient evidence to support a charge of financial fraud, but there have been control weaknesses historically, and the current accounting profit is very hard to read. If an economic downturn or a BTC downturn occurs while the AI/HPC transition also stumbles, the company can still survive, but shareholder returns would look very ugly.

Owner Earnings and Real Earnings Power

Analyzing RIOT with a Buffett-style Owner Earnings approach, the biggest difficulty is not the formula but that maintenance capital expenditure simply cannot be cleanly separated out by management. In the mining industry this is especially critical: miners depreciate quickly, and protocol halvings together with rising network difficulty shorten the "economic life" of machines, so much of what looks on the surface like "growth capex" is in fact, to a large degree, spending to maintain competitive position.

First the facts. In 2025 the company mined 5,686 BTC, with production value per BTC of about 101,350 dollars, for mining revenue value of about 576.3 million dollars; cash cost per BTC (excluding miner depreciation) of about 49,645 dollars, for total cash cost of about 282.3 million dollars; and all-in cost per BTC (including miner depreciation) of about 91,427 dollars, for total all-in cost of about 519.9 million dollars. This means that at the mine level alone, the 2025 mining business still had roughly 294 million dollars of cash gross profit before depreciation, and roughly 56.4 million dollars of gross profit after depreciation. The problem is that company-level SG&A, expansion spending, equity compensation, external-investment losses, and infrastructure construction quickly consume that thin profit.

Now Q1 2026. The company's total revenue for the quarter was 167.2 million dollars, of which Data Center revenue was 33.15 million dollars against a cost of 30.77 million dollars; Engineering revenue was 22.17 million dollars against a cost of 18.14 million dollars; and mining revenue was 111.9 million dollars against a cost of 86.76 million dollars. From a segment view, the data-center business already has scale, but is still far from the stage of proving "high margin, high cash return." The most conservative inference here is: do not equate the data-center narrative directly with a mature data-center cash cow.

I therefore do not give a falsely precise point estimate for Riot's Owner Earnings, but a range:

The logic behind this range is that Riot today cannot yet be treated as a machine that steadily generates cash, only as an enterprise with physical assets, BTC inventory, and transition options, but without yet-verified, high-quality Owner Earnings. For value investing, such an enterprise is attractive only when the price is very cheap; and the current price is not cheap.

Valuation, Margin of Safety, and Opportunity Cost

A Snapshot of the Current Valuation

As of May 20, 2026, RIOT traded at about 22.65 dollars. Market-data providers put its market cap at about 7.87 billion dollars; but the company's Q1 2026 disclosure of shares outstanding was 378.98 million shares, and multiplying that share count directly by the current price implies an equity value of about 8.58 billion dollars. This suggests the share-count basis in the market data may reflect a timing difference, so in valuation I place more weight on per-share value than on a single market-cap figure. Using Q1 2026 shareholders' equity of 2.395 billion dollars, book value is about 6.32 dollars per share, for a P/B of about 3.6x.

In addition, the company held about 15,680 BTC in Q1 2026; at the current BTC price of 77,388 dollars, the market value of the BTC inventory is roughly 1.213 billion dollars, or about 3.20 dollars per share. In other words, at the current price the market assigns a very substantial value to Riot's "non-BTC operating assets and future options." Using the reported share count, debt, cash, and BTC carrying value as a rough estimate, the implied value the market places on its "operating business itself" is still around 8 billion dollars. That is not a modest requirement.

Estimating Intrinsic Value

Method one: Owner Earnings discounting. Because current Owner Earnings are highly uncertain, I can only do scenario valuation rather than give a "precise value."

Note: the DCF above already incorporates the current net financial assets as part of the base value; but it does not pay an excessive option premium for the "not-yet-contracted forward AI/HPC narrative." That is also the biggest source of the gap between this estimate and the market price.

Method two: relative valuation. The biggest trouble in this industry is that most peers are unsuitable for traditional multiples. As of May 20, 2026, the trailing P/E of RIOT, MARA, and CIFR is negative; IREN's is positive but about 73x; CleanSpark has no meaningful P/E in the current data source. For Riot itself, 2025 EBITDA is negative and recent-year FCF is negative, so EV/EBITDA and P/FCF have no stable explanatory power; the most telling figure is in fact P/B of about 3.6x. This means that in this industry, the boundary between "cheap" and "not cheap" does not rest on traditional multiples but on whose assets can ultimately convert into stable contracted cash flow. On that count, Riot is still in the process of proving itself.

Method three: asset / liquidation value. Viewing RIOT very conservatively, valuing cash at 100%, BTC at book or a light discount, PP&E at a steep discount, and goodwill near 0, the liquidation value would land in roughly the 4–7 dollars per share range. Even with a milder treatment of BTC and land / data-center assets, it is hard to push the asset-based value anywhere near the current price. In other words, the current price rests not on a static asset cushion but on confidence in the success of the future conversion.

Combining the three methods, I arrive at the following ranges:

At the current price of 22.65 dollars, it sits roughly at the upper edge of my optimistic range, clearly above my fair range. Therefore:

These are not "precise bullseyes," but they answer a more fundamental question: if you treated yourself as someone acquiring the entire business, does the current price give you enough room for error? My answer is: not enough.

Margin of Safety and Opportunity Cost

In Riot's current valuation, the most fragile assumption is not the BTC price itself, but the market's default belief that it can smoothly convert a large amount of Texas power capacity into AI/HPC contracted cash flow at high returns. Once that assumption fails to hold, through delivery delays, excessive capital expenditure, slower-than-expected customer expansion, or unit returns below imagination, RIOT's valuation will revert to a "miner + asset package" frame; at that point, the fact that the current price lacks a margin of safety would be quickly exposed.

Compared with other opportunities, RIOT does not stand out. For ordinary long-term investors, a broad index like SPY represents a more diversified, more predictable equity asset; meanwhile the U.S. 10-year Treasury yield has recently been around 4.6%–4.7%, already setting a far-from-trivial opportunity-cost hurdle for risk capital. That is, RIOT's expected return must be meaningfully above this hurdle to justify bearing its industry, execution, and dilution risks. On my current valuation framework, it does not clear it. If I were allowed to hold only five assets for the long term, RIOT would not qualify for the portfolio.

Risk List, Counterarguments, and Final Judgment

Riot's most important risk is not "whether the share price will be volatile" but permanent loss of capital. The core risk chain runs like this: if the BTC price is not sufficiently strong over the long run, network difficulty keeps rising, and miner renewal sustains high capital intensity, while the AI/HPC transition fails to produce high-quality contracted revenue soon enough, then the company must either keep issuing equity and diluting, keep adding leverage, or sell BTC assets to sustain expansion. In that case, the high valuation shareholders pay when buying today could in the future be hit by the triple blow of "profits failing to materialize + dilution that does not stop + a fading narrative."

Looking more closely, competitive risk comes from the commodity nature of mining; technology-substitution risk comes from ASIC efficiency iteration and the constantly rising data-center specifications demanded by AI/HPC customers; regulatory risk comes from the power market, crypto-asset regulation, and local policy; financial risk comes from the 200 million dollar Coinbase loan, the rolling credit facility, and potential future financing; management risk comes from high equity incentives, historical internal-control problems, and the temptation to "grow scale rather than grow per-share value"; valuation risk comes from a current price that already embeds a far-from-low expectation of success; supply-chain risk comes from miner equipment and engineering delivery; and accounting risk comes from BTC fair value, derivatives, and high-volatility non-cash items that make the income statement very hard to read.

The strongest counterargument can be distilled into one sentence: RIOT is not "a good company unfairly punished," but more likely "a high-volatility asset layered with a high-expectation narrative, re-rated too far before its cash flow has been verified." Investors who are bearish typically see three things: first, the mining business has no pricing power; second, the AI/HPC story has not been validated by a large body of signed contracts and high margins; third, shareholders have repeatedly paid the price of dilution for expansion. I would admit my optimistic inferences were wrong if the following facts emerge: AMD's capacity expansion or new tenants fail to materialize for a long time; large negative free cash flow persists through 2026–2027; the company continues issuing equity at high cost during its "miner-to-infrastructure-platform" shift; and the return on invested capital per MW is clearly below what the market imagines.

Investment Checklist

The essential conclusion of this checklist is: by Buffett-style value-investing standards, RIOT is still far from "comfortable to buy."

Final Investment Conclusion

【Final Rating】 Watch

【One-Sentence Investment Thesis】 Riot holds noteworthy power / land / data-center transition options, but until its real Owner Earnings, return on capital, and shareholder-return discipline are verified, it still falls short of the buy threshold for long-term value investing.

【Core Bull Case】 First, Riot's Texas power and land assets are scarce, with Rockdale and Corsicana totaling 1.7 GW of approved capacity, providing the physical basis for the AI/HPC transition. Second, AMD has signed on and expanded to 50 MW, showing the data-center story is not pure PowerPoint. Third, the company still holds a large amount of BTC and substantial physical assets; it is not a "vapor company." Fourth, the decision to halt the Corsicana second-phase mining expansion and pivot to AI/HPC shows management has at least some capacity for strategic adjustment.

【Core Bear Case】 First, the mining business has no real pricing power and remains a commodity competition. Second, operating and free cash flow have been persistently weak or even negative in recent years. Third, share dilution is significant, and the capital-allocation record is dragged down by the Bitfarms episode. Fourth, the current valuation already capitalizes AI/HPC optionality in advance, but contracted cash flow and returns have not been sufficiently verified. Fifth, GAAP profit is heavily disturbed by fair value and non-cash items, making short-term profit unsuitable for "proving a great company."

【Key Assumptions】 For the investment to hold, at least the following conditions must be met: the AMD project delivers smoothly and keeps expanding; Corsicana / Rockdale can attract more high-quality AI/HPC tenants; the company stops relying on large-scale dilution to refill its coffers over the next two to three years; and even if mining is not the core profit source, it can at least keep cash flow from bleeding persistently.

【Fair Buy Price】 I believe the more reasonable buy range is 8–11 dollars per share. The basis is not a single multiple, but that within this range investors can at least obtain its asset floor, BTC inventory, and AI/HPC optionality at a price closer to fair intrinsic value, rather than paying in advance, at the current price, for a long stretch of unrealized future profit.

【Target Holding Period】 If fundamentals materialize in the future and the price is right, a horizon of 5–10 years would be appropriate; but at the current price, I would rather place it on a "high-quality watch list" than a "long-term core holdings list."

【Expected Annualized Return】 Working backward from the current price of about 22.65 dollars: if 10 years out it can realize only a conservative value of roughly 8–12 dollars, the annualized return is roughly -10% to -6%; if it realizes a neutral value of 16–18 dollars, the annualized return is roughly -3% to -2%; and even in a fairly optimistic case where value rises to 30–40 dollars in 10 years, the annualized return is only about 3% to 6%. This set of return / risk ratios does not stand out against the current 4.6%–4.7% 10-year Treasury opportunity cost.

【Maximum Loss Risk】 In the worst case, if BTC stays depressed, network difficulty keeps rising, the AI/HPC transition stalls, and the company keeps diluting or is forced to sell coins or add leverage, the share price could drift closer to asset / liquidation value, around 4–7 dollars per share. Relative to the current price, this means a roughly 70%–80% permanent loss of capital is not unimaginable.

【Tracking Metrics】 What should be tracked going forward is not the share price but: signed AI/HPC load in MW, invested capital and expected return per MW, Data Center segment gross margin / EBITDA, cash cost per mined BTC, average operating hashrate and downtime, operating cash flow, economic FCF, changes in share count, changes in BTC holdings, and the level of debt and collateralized BTC. These metrics will tell you whether the investment logic is improving earlier than "short-term EPS" will.

【Signals That Trigger Reassessment】 If a large new AI/HPC customer lands, the Data Center segment's margin improves markedly, and the company stops significant dilution and begins to show stable Owner Earnings, reassess; conversely, if large negative free cash flow persists through 2026–2027, AMD's capacity expansion stalls, further high-cost equity issuance occurs, or management once again makes a major capital-allocation error, reassessment is also required, biased toward lowering the conclusion.

【Final Recommendation】 Soberly put, RIOT is worth watching, but at present it is not worth a heavy buy on value-investing logic. If you are a long-term investor with a balanced risk appetite, the best move is not to chase the upper bound of this company's narrative, but to wait for one of two things to happen first: either the price falls clearly to a level with more margin of safety, or the data-center transition is first verified by persistent cash flow in the statements. Until either occurs, placing it in the "Watch" column rather than the "Buy" column is both disciplined and more consistent with the principle of long-term capital preservation.

Open Questions and Limitations

This report relies as far as possible on the company's annual reports, quarterly reports, proxy statements, and official announcements, but several pieces of information still cannot be confirmed with false precision: first, the precise definition of maintenance capital expenditure; second, the complete unit-economics model of the AI/HPC projects; and third, the comparable financial bases of some peers at the same point in time. The Owner Earnings and DCF in this report should therefore be understood as conservative range inferences, not a mechanical "target-price formula." This does not change the core judgment: the current price lacks a margin of safety.

This report is based on public information and does not constitute investment advice. Markets carry risk; invest with caution.