Bitdeer Technologies: A Deep-Value Investment Study

Business understandability score: 3/5.

Industry and Competitive Landscape

Bitcoin mining is fundamentally not an easy industry to win over the long run. Long-term demand does exist: as long as the Bitcoin network exists, miners need hash rate and power. But the profit pool is not stable, and returns are continually eroded by three forces: block-reward halvings, rising network hash rate, and rapid hardware depreciation. In April 2024, Bitcoin completed its fourth halving, cutting the block reward from 6.25 BTC to 3.125 BTC. Meanwhile, Cambridge's digital-mining research shows Bitcoin mining consuming roughly 138 TWh of electricity annually, with the United States already the largest global mining hub at about 75.4% of the sample. This shows the industry is still expanding, but expansion does not equal stable excess profit.

AI infrastructure is a more attractive arena, but Bitdeer is far from a strong player on it. The International Energy Agency expects global data-center electricity use to rise to roughly 945 TWh by 2030, a sharp increase from today, with AI a core driver. This direction is of course more compelling than pure mining. The problem is that Bitdeer states it plainly: in AI infrastructure and AI cloud, it faces hyperscale compute clouds, specialized AI infrastructure providers, and established data-center operators that hold stronger capital, technology, supply chains, and customer relationships, and the company cannot guarantee that it can establish a meaningful competitive position.

Competitors and industry position: The main competitors Bitdeer names in its annual report include MARA, RIOT, CleanSpark, IREN, Cipher, Core Scientific, CoreWeave, and Nebius. The company also concedes that many rivals are larger, have stronger brands, are better capitalized, and hold deeper supplier relationships. Put differently, Bitdeer is not an outright leader in "mining," nor has it entered the top tier in "AI infrastructure." Its real position looks more like this: it wants to capture profit upstream in the mining chain (ASIC/rigs) while also commanding a valuation premium downstream (data centers/AI), yet it has secured a moat on neither side.

Whether the industry profit pool is concentrated: In mining, profit concentrates more in the participants with the lowest power cost, highest equipment efficiency, and best financing ability; in AI infrastructure, profit concentrates in companies that have power, have land, have GPU supply, and can sign customers. Bitdeer holds some of these conditions, but none is yet strong enough to be "hard for others to catch." From a long-term owner's perspective, this looks more like "a hard-working company in a poor industry, plus an attractive but unproven AI option" than "a good company in a good industry."

Industry attractiveness score: 2/5.

Moat

If you take it apart strictly along a Buffett-style framework, Bitdeer's moat is locally present but weak overall.

Brand advantage: weak. In the sense that matters to end consumers, Bitdeer has no Coca-Cola-style brand; within the industry it has some recognition in cloud hash rate and mine-site operations, but that is not enough to create a strong premium or defense. On the contrary, the company itself acknowledges that many competitors have stronger brands.

Cost advantage: early signs, but not proven durable. The company owns mine sites and data centers across multiple locations and is pushing in-house ASICs and SEALMINER mass production; if rig efficiency, yield, and supply cadence deliver, a per-unit hash-rate cost advantage could indeed form. The company also discloses operating techniques such as "hash rate slicing" and "hash rate scheduling," and management emphasizes its first-mover experience in hosting and rig management. The catch is that these advantages still have to pass two tests: first, whether it can genuinely secure scale advantages in the supply chain; second, whether it can withstand rising Bitcoin difficulty and tight GPU supply.

Scale advantage: moderate. Scale matters in mining, but industry leaders do not earn from "scale" alone; they earn more from "cost of capital plus power cost plus equipment efficiency." Bitdeer has some scale, but against MARA, RIOT, CleanSpark, and IREN, that scale carries no overwhelming edge; and in AI infrastructure, it is even less of a scale leader.

Network effects: almost none. Cloud-hash-rate platforms and hosting platforms do not inherently carry strong network effects; more users do not raise platform value exponentially the way a social network does. Even with some ecosystem stickiness, it falls well short of "the more users, the harder to replace."

Switching costs: low to moderate. Mine-site hosting and hash-rate purchases carry some technical migration cost, but it is not unbearable; more decisively, customers ultimately look at price, time-to-deploy, stability, and power conditions. That is, Bitdeer's customer stickiness is more "service stickiness" than "structural lock-in."

Patents, licenses, data, and operating capability: moderate, but not enough alone to form a strong moat. As of the end of 2025, the company held 45 registered patents, 116 patent applications, and 304 registered trademarks. This shows it is not an empty-concept shell but a company with some technical accumulation. The problem is that where rigs and AI infrastructure truly decide the outcome, what matters is often not the patent count but manufacturing capability, foundry/packaging-and-test relationships, equipment supply, speed to power-on, and customer contracting. These are precisely the parts the company itself admits, in its risk disclosures, remain constrained by external resources.

Capital-allocation ability: not a moat at present. This is where I am least willing to give a high score. Buying back stock is not bad in itself, but if a company simultaneously dilutes/levers up on a larger scale through convertibles, ATM offerings, or debt financing, the buyback is hard to call an allocation that "raises intrinsic value per share" and looks more like a tactical move. Bitdeer did repurchase shares in 2025, but in the same year it also issued large amounts of stock to settle convertibles and markedly pushed up total share count and debt.

Moat assessment: On the whole, I see its moat as neither clearly widening nor collapsing, but in a state of "narrow and dependent on execution." If the AI pivot delivers, it could widen; if the rig business and AI business come under pressure at once, the moat would instead contract. Replicating it would take considerable capital and time, but for rivals with lower capital costs and stronger supply-chain relationships, it is not impossible to replicate.

Pricing power in inflation: almost none in mining; limited in hosting; possible in AI if it secures scarce power and GPU resources. Ability to stay profitable in a downturn: not at present. Whether past high margins were structural or a cyclical/accounting windfall: closer to cyclical and accounting noise, not a proven structural advantage.

Moat strength score: 2/5.

Management and Capital Allocation

Start with the positives. Jihan Wu is the company's founder and has served as CEO since March 2024; as of April 2026 he held roughly 69.5% of voting power, meaning founder control is very strong. For an early-stage or transition-stage company, this kind of control can sometimes ensure strategic continuity rather than being pulled around by the short-term market. Management has also disclosed competition, regulatory, supply-chain, power, and AI-expansion risks fairly fully in its filings.

But from a long-term outside-shareholder view, there are more concerns. First, control and economic interest are not aligned. Jihan Wu maintains very high voting power through Class V shares, while common shareholders have very limited ability to check capital allocation. Second, Tether-related entities hold roughly 19.3% of Class A, so the shareholder base is itself highly concentrated. Third, the company began using related-party financing heavily in 2025: year-end borrowings from BIT Group stood at about 522 million dollars, at a rate of 9% plus a market benchmark rate, collateralized by rigs, inventory, data-center assets, and the like; by Q1 2026, both the short-term and long-term balances of related-party borrowings remained large. For outside shareholders, this structure is not necessarily unreasonable, but it markedly raises governance complexity.

Whether capital allocation is rational: Here one must separate "reasonable expansion" from "per-share-value-friendly." Bitdeer sharply increased R&D in 2025, advancing in-house ASICs and SEALMINER, which is logical strategically; but at the execution level, 2025 saw both large-scale buybacks and large-scale share issuance / convertible settlement at the same time. The annual report shows the company repurchased about 4.965 million Class A shares in 2025 for roughly 56 million dollars; yet in the same year it issued 11.048 million and 10.661 million Class A shares for 2024 convertible settlement, buybacks, and related matters. In other words, the buybacks did not actually offset the dilution. For value investors most sensitive to "intrinsic value per share," this is not an ideal capital-allocation picture.

How cash is used: Looking at the 2025 and Q1 2026 cash flows, cash went mainly to data-center construction, rig/chip/raw-material procurement, digital-asset-related flows, debt service and refinancing, and a small amount of buybacks; there were no dividends, and the cash truly free to distribute to shareholders is not evident. In Q1 2026, the company's net financing cash inflow was about 353 million dollars, but operating cash flow was a net outflow of about 347 million dollars, and the investing side put in nearly 94 million dollars of capex. This looks more like a "growth propped up by financing" phase than a "growth that feeds itself" phase.

Whether M&A and investments create value: I am unwilling to give an affirmative verdict. The company still carries goodwill and intangibles on its books; at the end of 2025, goodwill was about 35.82 million dollars and intangibles about 93.43 million dollars, while the impairment testing of part of its asset groups already relies clearly on a future AI pivot. Especially worth watching is the valuation in the annual report of the asset group tied to the Tydal data center, which assumes that, after the shift to AI infrastructure, budgeted EBIT can grow at an average annual rate of 80.2%. This assumption is not impossible, but it is quite aggressive. Should the AI retrofit run slower than expected, utilization fall short, or customer contracting disappoint, the gap between book value and true economic value would widen quickly.

Management and capital-allocation score: 2/5.

Financial Quality, Owner Earnings, and Valuation

First, the key financials. In the table, the 2022-2023 data come mainly from the 2023/2024 20-F, the 2024-2025 data from the 2026 20-F, and the Q1 2026 update from the company's 6-K / quarterly filing. Some ratios in the table are my own calculations based on disclosed figures.

• Here I treat "purchases of property, equipment, and intangibles" plus "purchases of mining rigs / rig equipment" combined as an approximation of capex. ** Rough free cash flow = operating cash flow minus the combined capex above; it is not the company's measure, but it is enough to show the direction of capex and cash burn.

Key conclusion one: reported profit does not equal real cash profit. In 2025 the company looked like it "turned profitable," with net income of about 65.6 million dollars; but in the same year "other net gains" reached 365 million dollars, of which the fair-value change in derivative liabilities alone contributed about 445 million dollars. Strip out these fair-value swings, which have little to do with operating quality, and 2025 core operations show no reassuring inflection point; operating cash flow that year was in fact -1.739 billion dollars. This is why I regard the 2025 P/E as carrying almost no analytical meaning.

Key conclusion two: growth depends heavily on capital input, and right now the feature is "the more it grows, the more cash it needs." At the end of 2025, the company held only 149 million dollars of cash, while prepayments and other assets were close to 698 million dollars and inventory about 252 million dollars; by Q1 2026, prepayments fell to 341 million dollars, but inventory climbed further to 613 million dollars. The company explains this is mainly due to the increase in wafer, chip, work-in-process, and finished-goods inventory brought on by SEALMINER mass production at scale. In other words, the company is using real cash to stack capacity, equipment, and raw materials, and when those inputs turn into high-quality cash recovery remains unknown.

Key conclusion three: the balance sheet is not robust. As of Q1 2026, the company disclosed borrowings of about 1.9 billion dollars, while cash, cash equivalents, and restricted cash totaled about 298 million dollars; even counting digital assets and digital-asset receivables, the liquidity buffer is hardly ample. More troubling, the assets here include large amounts of inventory, prepayments, specialized rigs, data-center assets, and digital-asset-related receivables, which look valuable in an up-cycle but whose recovery value falls sharply in a down-cycle.

Key conclusion four: per-share value is continually challenged by dilution pressure. Basic weighted-average shares in 2023, 2024, and 2025 were roughly 110 million, 137 million, and 205 million; as of April 21, 2026, common shares outstanding totaled about 243 million, with a further roughly 8.469 million Class A shares reserved for equity incentives. Even with occasional buybacks, the company is overall still in a reality of "share-count expansion."

On ROE, ROIC, interest coverage, and P/FCF: None of these metrics looks good for Bitdeer, and some are severely distorted by accounting swings. The 2024 statements showed a huge loss, with ROE/ROA sharply negative; 2025 ROE looked improved on the surface, but was driven mainly by non-cash fair-value gains. Q1 2026 returned to an operating loss, with net interest expense of about 29.52 million dollars against a clear operating-level loss, so current interest coverage is not safe. P/FCF is essentially unusable, because free cash flow has been deeply negative for several years.

Owner-earnings estimate: I take a more conservative, "owner" rather than "trader" view. On the facts, 2025 net income was about 65.6 million dollars, and adding back depreciation and amortization gives about 146.7 million dollars; but I do not treat all share-based compensation as a cost-free add-back, and given the rapid economic depreciation of rigs, the cyclical replacement, and data-center equipment maintenance, maintenance capex is unlikely to be below depreciation and may be higher. Factoring in the inventory/prepayment working capital tied up to maintain scale, I judge that under the current run-rate of 2025 and 2026, conservative owner earnings remain negative, falling roughly in the -150 million to -300 million dollar range, with a midpoint of about -225 million dollars. This is an inference, not a company-disclosed figure, but it is closer than net income to the economic profit shareholders can actually take home.

My bottom line on owner earnings: This company has not yet proven it can generate, over the long run and stably, free cash "distributable to owners." What it must first prove is that heavy capital input can become high-return cash flow in the future, rather than continuing to turn into inventory, prepayments, and refinancing. Until that genuinely happens, I will not award a "good business" high score on the strength of a single year of accounting profit.

Valuation method one: discounted owner earnings. I use the current economic-interest share count of about 243.3 million shares as the denominator and run conservative estimates across three scenarios. The cash-flow path here is my inference, not company guidance. The discount rates are 14% / 12% / 11% and the terminal growth rates 2% / 3% / 4%, to reflect this company's risk being clearly above that of an ordinary mature enterprise. The results:

The relationship between these ranges and the current price is this: 12.83 dollars is already above what I see as the "fair-value midpoint," but not yet high enough to break entirely free of the optimistic scenario. That is, the market is already paying a fair part of its expectation for "AI plus in-house rigs plus scale delivery."

Valuation method two: relative valuation. In the comparison below, for BTDR I use the equity value I calculated myself on the "economic interest of all common shares"; the other companies use the market cap from current market-data tools. Traditional P/E, EV/EBITDA, and P/FCF are broadly unstable across this group, so I weight P/B and P/S more heavily, since they are relatively less affected by fair-value swings. The 2025/latest shareholders' equity and revenue come from each company's official annual report or results disclosure.

This table makes one important point clear: on a long-term owner basis, Bitdeer is not cheap. It is not an "asset stock the market is ignoring." On the contrary, once the economic interest of all common shares is counted in, BTDR's P/B is even higher than MARA, CLSK, and RIOT, and below only IREN with its obvious AI-mania premium; yet its cash-conversion ability, debt pressure, and governance complexity are no better.

Valuation method three: asset / liquidation value. On the surface this company has many assets: cash, digital assets, inventory, prepayments, rigs, data centers, real property, and projects under construction. The problem is that a large share of the assets is highly industry-specific. Rigs and data-center equipment often discount quickly when the industry faces headwinds, while inventory and prepayments face supply-chain, yield, and demand-delivery risk directly. So if you apply a conservative discount for a liquidation valuation, existing liabilities would consume a sizable part of asset value; I judge that the "safety floor" the asset method gives BTDR is not strong, with conservative liquidation equity value supporting only roughly a 0 to 5 dollars per share range. That is, the asset method offers no clear protection for the current price of 12.83 dollars.

Margin-of-safety assessment: The most fragile assumption behind the current price is that the AI retrofit and in-house-rig spending will ultimately become high-return cash flow. If that assumption fails, the market will re-price Bitdeer as a high-leverage, highly cyclical, weak-cash-flow mining-infrastructure platform; at that point the valuation midpoint would very likely fall back to my "base range" or even "conservative range." So the margin of safety is not sufficient.

My price-band assessment:

• Conservative intrinsic value range: 2 to 5 dollars

• Base intrinsic value range: 7 to 11 dollars

• Optimistic intrinsic value range: 15 to 22 dollars

• Ideal buy price range: 5 to 7 dollars

• Acceptable holding price range: 8 to 12 dollars (provided a position is already established and you explicitly accept the AI-option nature)

• Clearly overvalued price range: above 15 dollars, unless a large AI contract, markedly improved cash flow, and deleveraging evidence subsequently appear.

Risks, Comparisons, Checklist, and Final Conclusion

The most important risk is not short-term price volatility but permanent loss of capital. I care most about the following categories:

• Competition risk: no stable pricing power on the mining side, and on the AI side it faces stronger, wealthier rivals with more customer resources.

• Technology-substitution and execution risk: if SEALMINER's mass production, efficiency, yield, and cost curve fall short of expectations, the pre-built inventory and prepayments will quickly become sunk costs.

• Regulatory risk: crypto-asset rules keep evolving, and AI data centers also face rising scrutiny over energy use, environmental impact, and community review. Bitdeer's own annual report already flags tightening U.S. and multi-state scrutiny of AI data centers.

• Financial-leverage and refinancing risk: by Q1 2026, the company's borrowings were about 1.9 billion dollars, while operating cash flow remained a large outflow. If capital markets tighten, the pressure would be very real.

• Supply-chain risk: single power supplier per site, limited GPU suppliers, and a lack of long-term capacity-guarantee agreements.

• Accounting-noise risk: profit is easily affected by digital assets, derivatives, fair value, and debt restructuring, which can lead investors to mistake accounting gains for economic gains.

• Governance and related-party risk: dual-class structure, high founder voting power, and large related-party financing balances.

The strongest bear case is in fact straightforward: Bitdeer is neither a "cheap AI stock" nor a "stable mining stock"; it is more like a capital-intensive platform still burning cash to reshape its business model. What a bear or short is most likely to see: 2025 profit was not real, 2026 cash flow is deteriorating again, the AI pivot is just a story, and debt and dilution are the reality. If, over the next 12 to 24 months, the company cannot turn the AI business into substantive contracts, deliver verifiable gross-margin improvement from in-house rigs, and meaningfully narrow the operating cash outflow, the current valuation will be hard to defend.

Which facts would overturn the current investment judgment: If the following occur, I would admit today's cautious judgment is wrong and needs to be revised up: First, the AI data-center retrofit (especially Tydal) lands on schedule and forms multi-year, high-quality contracted revenue; second, SEALMINER's mass production not only ramps in volume but also matches mainstream rivals on energy efficiency and unit economics; third, operating cash flow improves markedly, inventory and prepayments stop consuming cash, and the pace of debt/share-count expansion slows. Conversely, as long as two of these three fail to deliver over the long run, the stock is more likely to fall into the "capital-intensive plus storytelling plus dilution" trap.

Comparison with other opportunities: If I had to compare within the same basket today, Bitdeer is not clearly superior to simpler alternatives. MARA had 2025 revenue of about 907 million dollars, total equity of about 3.48 billion dollars, total BTC holdings of about 53,822 coins, and roughly 1.9GW of capacity with 66.4 EH/s of energized hash rate; RIOT had 2025 revenue of about 647 million dollars and stresses its near-2GW power portfolio with its AMD data-center lease now operating; CleanSpark had fiscal-2025 revenue of about 766 million dollars and operating hash rate above 50 EH/s; and SPY, as the S&P 500 ETF, currently trades at about 733.73 dollars, representing diversified U.S. large-cap exposure; in addition, the U.S. 10-year Treasury yield today is about 4.66% to 4.69%. Under this comparison, BTDR's neutral-return outlook offers no clear compensation for its extra risk.

If I could hold only 5 assets, does it qualify for the portfolio? At the current price and on current information, my answer is: no. The reason is not that it has no upside, but that its upside depends excessively on several unproven key assumptions; high-quality long-term portfolios usually lean toward assets that "let you hold comfortably even without the optimistic scenario."

Investment Checklist

Open questions / limitations: The public materials I can access are enough to judge that the stock is "currently not cheap, weak on cash flow, with AI still an option," but several items still need follow-up tracking: the term and per-MW/per-GPU return of the latest AI customer contracts, SEALMINER's mass-production yield and external-sales share, whether related-party financing keeps expanding, and whether the company can pull second-half-2026 operating cash flow back into a clearly improving range. Public materials do not yet provide a sufficiently mature data loop on these points.

【Final Rating】 Watch

【One-line investment thesis】 Bitdeer today looks more like "a high-capex crypto-infrastructure platform plus an unproven AI option" than a long-term value stock with stable owner earnings and a clear margin of safety.

【Core bull case】

• The company is not a pure miner; its reach extends into ASICs, in-house rigs, data centers, and AI infrastructure, with large valuation upside if it succeeds.

• Founder control is strong and strategic execution has continuity; the company is genuinely spending on R&D and building capacity rather than staying at the slogan stage.

• If AI utilization, customer contracting, and SEALMINER unit economics deliver, the current bearish cash flow may be only the eve of a turning point.

【Core bear case】

• Truly distributable cash flow is negative, and both operating and free cash flow in 2025-2026 are poor.

• 2025 profit quality is low, heavily affected by fair-value and derivative accounting, and the P/E is strongly misleading.

• Debt, inventory, prepayment, and dilution pressures are all considerable, and the AI business is still unproven.

【Key assumptions】

• AI data-center retrofits such as Tydal proceed on schedule and form sizable, renewable revenue;

• SEALMINER's mass production, yield, and energy efficiency reach a competitive level;

• Operating cash flow improves markedly over the next 4 to 8 quarters rather than continuing to fill holes with financing.

【Fair Buy Price】 I would prefer to consider building a watch position or small position in the 5 to 7 dollar range, because that leaves a sizable discount to the lower bound of my "fair value range" and better covers the risks of AI execution falling short, mining-cycle swings, and subsequent dilution.

【Target holding period】 If you do buy, you should view it as at least a 3-to-5-year-plus project-style hold, where the real story is the AI pivot and in-house-rig delivery, not quarterly swings. For pure value holders of 10 years or more, I would not for now treat it as a core position.

【Expected annualized return】

• Conservative scenario: about -12% to -6%;

• Neutral scenario: about 1% to 5%;

• Optimistic scenario: about 10% to 15%. This range reflects that the current price already embeds part of the success expectation, so only in the optimistic scenario does the return clearly beat the risk-free yield and the index alternative.

【Maximum loss risk】 In the worst case, if the AI pivot fails, rig mass production misses targets, the financing environment worsens, and the Bitcoin price and difficulty deal a double blow, the balance sheet could force the company into continued large refinancing or dilution, and the share price returning to my conservative valuation range would not be absurd; from the current price, a permanent drawdown of more than 70% is not unimaginable.

【Tracking metrics】 The items most worth tracking going forward:

• AI data-center contracted capacity, utilization, and per-MW/per-GPU revenue;

• SEALMINER shipment volume, energy efficiency, gross margin, and external-sales share;

• Self-mining unit cost and downtime/power-on efficiency;

• Whether operating and free cash flow converge;

• Changes in inventory, prepayments, and receivables;

• Related-party borrowing balances and financing costs;

• Total share count and potential dilution;

• Capex intensity;

• The pace of disposing of digital assets and digital-asset receivables;

• Whether any major impairment or restatement occurs.

【Signals that trigger a re-evaluation】 If any of the following occur, I would immediately recompute the valuation:

• The AI business secures a substantive long-term contract;

• Operating cash flow improves markedly for two or more consecutive quarters;

• SEALMINER unit economics are validated by customers or the financials;

• Debt continues to rise quickly;

• Inventory and prepayments expand markedly again;

• A large asset impairment or larger-than-expected dilution occurs.

【Final recommendation】 Put coolly, Bitdeer is worth tracking, but not worth a heavy position at the current price on "Buffett-style high-certainty" logic. If you think like a long-term business owner, the most important thing is not to bet on "whether it will tell a bigger AI story," but to wait for it to prove three things: first, that profit can become cash; second, that expansion raises rather than dilutes per-share value; third, that the moat is not stacked up by financing. Until these three appear, the most fitting label I give it remains: Watch, not Buy.

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