Canaan Inc. — Zen Horizon Framework Deep Dive

• Business quality: a bad business, moat 2/5. Bottom of the three rig giants (share <5% on a hashrate basis), an energy efficiency about one generation behind Bitmain (its own water-cooled line nearly two generations behind), no pricing power over the coin price (single-digit or even negative gross margin), four consecutive years of losses totaling about USD 963 million, and a history of value destruction (−96% since IPO, −99% from peak, with never a reverse split). The former second growth curve — AI chips (Kendryte) — was shut down in June 2025. This is the fundamental reason for "Avoid" rather than "Watch" — it is not a good business.

• Essence: a pure Bitcoin high-beta speculative instrument once the AI narrative is stripped away. With the Kendryte AI line exited, revenue is ≈100% from "selling rigs + self-mining," both of which are amplifiers of the Bitcoin price. It has no stable cash flow that can ride through the cycle; the share price is a leveraged mapping of the coin price — these are the attributes of a speculative chip, not those of an investable security.

• Survival risk: death spiral × delisting clock × continuous dilution. Bitcoin's current price is about USD 61,900, −51% from the October 2025 high; listed miners' cash cost is USD 80,000–90,000, with most mining at a loss → frozen rig demand; the company keeps itself alive by drawing down via the ATM (cap USD 270 million), with Class A shares already expanded about 5.6x; the Nasdaq delisting deadline is 2026-07-13, and getting back to USD 1 requires +180% — having never done a reverse split historically, it will most likely be forced to reverse-split to save itself. The three are tightly interlocked — a survival-level tail risk.

• Price position: the cheapest = the least trusted. P/S 0.52x and P/B 0.70x are the lowest among peers; the mirror-image name Bitdeer (likewise self-developed ASIC + self-mining + AI cloud) has a market cap 16x its own, with smaller TTM revenue yet a far higher market cap. Canaan's TTM revenue (about 510 million) is larger than most comparables, yet its market cap is only 1/45–1/73 of theirs — the market is not overlooking it; it disbelieves that Canaan can turn revenue into sustainable profit, and disbelieves it can safely survive delisting and dilution.

Rating: Avoid. This draws a clear line from the good businesses that warrant "Watch" (share eroding but still the leader): this case stacks a bad business + high speculation + survival risk. The reason it is not given the most aggressive "Sell": equity is still positive (about USD 382 million), debt is very light, HFCAA risk has been removed, most bad news is already priced in, and going to zero is not the base case — it still retains speculative residual value as a deep-cyclical bullish option on Bitcoin. But it is not a company worthy of entrusting capital to, and rational investors should stay away. A symbolic ideal buy of ≤ USD 0.20 (close to liquidation residual value, and requiring listing-survival confirmation + the start of a coin-price bull market simultaneously); at USD 0.3581 there is no margin of safety.

II. Company Profile

2.1 What it actually is: the "shovel seller" at the very top of the crypto supply chain

Laying out the entire Bitcoin supply chain, Canaan sits at the very top: the base layer is the Bitcoin network itself → upstream is rig / hashrate hardware manufacturing (where Canaan sits) → midstream are the miners who buy rigs to mine coins (MARA / RIOT / CleanSpark, etc.) → downstream are exchanges (Coinbase), payments, and stablecoins.

The "shovel seller" logic: in a gold rush, the surest player is not the prospector but the one selling the shovels — in theory, no matter which miner strikes gold, they must first buy rigs (shovels) from a maker like Canaan. This is the theoretical appeal of Canaan's business model. But the reality is that this "shovel-selling" business has three brutal characteristics that make it far less attractive than the "shovel seller" narrative:

• The shovels are highly cyclical hardware with no pricing power. Rig prices swing violently with the price of Bitcoin: coin price up → miners scramble for rigs, volume and price both rise; coin price down → rigs go unsold, prices are cut to clear inventory. Canaan has no say over the end price; it is a naked price taker.

• Canaan's shovels are worse than rivals'. On the one hard metric of hashrate efficiency (J/TH, power per terahash), Canaan is systematically behind Bitmain (see the moat in Section V).

• Canaan later went mining itself too. To absorb capacity and smooth the cycle, Canaan in recent years took up a self-mining business (Q1'26 self-mining revenue of USD 19.1 million, output of 257 BTC, hashrate of about 11 EH/s). This turned it from a pure "shovel seller" into a "shovel seller + miner," but it also means it directly bears the dual risk of a falling coin price + mining losses.

2.2 Business segments: both legs planted in Bitcoin

Source: Canaan Q1'26 results 6-K. Key conclusion: both segments are amplifiers of the Bitcoin price, with no stable revenue that can ride through the cycle. This is precisely the opposite of the "rain or shine" imagery in the "shovel seller" narrative.

2.3 The former second growth curve is dead: the AI-chip narrative is bankrupt

Canaan once heavily promoted its edge AI-chip business (Kendryte, a RISC-V-architecture AI SoC) as a "de-Bitcoinization" second growth curve, and it was at one point the key narrative supporting its valuation imagination. But the reality is brutal:

• Kendryte contributed only about USD 0.9 million of revenue for all of FY2024, yet chronically consumed about 15% of the company's operating expenses — a textbook "burns cash, makes no money" business.

• The company already formally announced its exit from / divestment of the AI-chip business on 2025-06-23, concentrating resources on Bitcoin rigs and self-mining.

This means the "Canaan is an AI concept stock" narrative is bankrupt. An investor buying Canaan today is buying 100% pure Bitcoin beta, with no AI-story discount protection or imaginative upside remaining. All subsequent analysis in this report is based on the qualitative judgment of a "pure Bitcoin name."

III. Vertical Analysis (history and share-price record)

3.1 A history of value destruction: from USD 9 to USD 0.36, 96%–99% evaporated in six years

To understand Canaan, you must first confront its harrowing share-price record — this is the single most important chart in the case.

• 2019-11-21: IPO'd on Nasdaq at USD 9 / ADS, billed as "the world's first blockchain stock" and "the world's first Bitcoin mining stock."

• Early 2021: Riding the Bitcoin bull market (BTC first breaking USD 60,000), the stock surged to an all-time peak of about USD 39.

• 2021 → 2026: What followed was a five-year one-way collapse, through the 2022 crypto winter, the 2023 industry shakeout, the 2024 halving, and the back-and-forth of 2025, closing at USD 0.3581 on 2026-06-05.

• Cumulative drawdown: −96% from the IPO price, −99% from the peak, near the 52-week low of 0.358 and the all-time low since listing.

The most critical detail: Canaan has never executed a reverse split in six years. This means that 96%–99% drawdown is a genuine, accounting-unmodified destruction of wealth — not a "low price produced by splitting shares," but a literal burning of 96 cents of every long-term shareholder's principal. The best definition of a value-destruction machine is "long-term, irreversible, unmodified destruction of shareholder value," and Canaan fits it perfectly.

3.2 The revenue roller coaster: completely dragged along by the Bitcoin cycle

The violent swings in revenue (in USD terms) are the most direct evidence of "highly cyclical, no pricing power":

† FY2022 official figure is USD 634.9 million; the data vendor stockanalysis shows about 651.5 million due to currency-translation conventions — this report uses the official figure.

That a company's revenue can swing +1041% in a single year, and also −67%, is itself proof that it has almost no control over its own fate. This is a textbook "company sitting on the cycle," not a "company riding the cycle."

3.3 Earnings record: four consecutive years of losses, cumulatively about USD 963 million

• The only two profitable years were 2021–2022 (bull-market dividend), of which FY2022 net income was about +USD 69.9 million.

• Continuous losses from FY2023: FY2023, FY2024, FY2025, and Q1'26 were all loss-making, with the four loss periods cumulating a net loss of about USD 963 million.

• Clarifying the convention (important): This 963 million is the "sum of net losses across four loss-making years"; whereas the balance-sheet "accumulated deficit" is about USD 749.5 million — the latter is lower because the 2021–2022 profits partially offset subsequent losses within retained earnings. Both figures are correct, but the conventions differ; they must not be conflated or added together.

Whichever convention is used, the conclusion is the same: over the past four years Canaan has destroyed value on a scale approaching USD 1 billion, with no path to an organic turnaround in sight.

IV. Financial Review

4.1 Q1'26 (as of 2026-03-31): the bleeding is stark

Source: Canaan Q1'26 results 6-K.

4.2 Look at operating loss, not GAAP net loss (P/E and loss are distorted by one-offs)

Q1'26's GAAP net loss of 88.7 million is about 34.4 million larger than the operating loss of 54.3 million, with the gap coming almost entirely from non-operating items unrelated to the core business's ability to generate cash:

• Crypto-asset fair-value change: −24.9 million (coin price fell; unrealized loss on held BTC/ETH)

• Derivatives / other financial instruments: −16.0 million

• Inventory write-down: +24.5 million (charged to cost, depressing gross profit)

Significance for judgment: Using the GAAP net loss to compute "price-to-sales / loss multiples" would overstate the degree of bleeding; but conversely, these non-operating items also reveal the fact that its balance sheet is deeply hostage to the coin price (unrealized coin losses hit net income directly). This report uniformly uses the operating loss of 54.3 million to gauge core-business cash generation, and "equity and cash" to gauge survivability.

4.3 Cash and survival: bleeding vs. life support

• Cash fell from 80.8 million (start of year) to 43.5 million (end of Q1'26), a single-quarter net outflow of about 37.3 million; Q2 guidance points to further decline — at this rate of bleeding, the cash is only enough to last a few quarters.

• Life support comes from ATM issuance: The company has an ATM (At-The-Market) issuance program of about USD 270 million outstanding, allowing it to sell new shares into the secondary market for cash at any time — this is the source of its ability to keep "buying time," and the root of continuous shareholder dilution.

• Debt is very light: Interest-bearing debt is about USD 54.5 million, with no convertible bonds — this is one positive factor keeping it "not going to zero immediately" (it will not be forced into bankruptcy by creditors), but the price is shifting all the survival pressure onto equity dilution.

4.4 Coin-hoarding: the vast gap between the headline figure and the actual holdings

Canaan has in recent years copied other miners with a "Bitcoin treasury," but its disclosure conventions need to be unpacked:

• Headline figure: holds 1,807.60 BTC + 3,951.53 ETH.

• Actually self-owned: truly unencumbered, self-owned coins are only about 802.6 BTC; there are also 1,005 receivable (of which 905 are already pledged for collateralized loans and 100 are locked in fixed-term wealth products), plus 63.4 of customer deposits (which do not belong to the company).

• Physical total is about 1,871.0 BTC; on the financials, crypto-asset fair value is about USD 66.2 million + crypto-receivable fair value is about USD 67.0 million.

Significance for judgment: The headline "holds 1,807.60 BTC" gives the illusion of "a thick cushion," but truly freely usable, unpledged, unencumbered self-owned coins are less than half of that. In a death-spiral scenario, these pledged coins could be passively liquidated by margin calls triggered by a falling coin price — coin-hoarding is not a safety cushion; it may instead be an amplifier.

V. Moat: narrow and eroding (2/5)

Canaan's composite moat score is 2/5 (narrow) — this is the technical core of the "bad business" judgment. Broken down item by item:

5.1 Technology: catching up, and systematically behind (the most fatal)

The one hard metric in rig competition is hashrate efficiency (J/TH, power per terahash, lower is better) — it directly determines a miner's electricity cost and payback period. On this metric:

• Air-cooled flagship comparison: Canaan's A16XP at 12.8 J/TH is about one generation behind Bitmain's S23 Hydro at 9.5 J/TH.

• The water-cooled comparison is worse: Canaan is not without water-cooled products (correcting a common misconception), but its own water-cooled line (such as the A1566HA) has an efficiency of about 16.8 J/TH, nearly two generations behind Bitmain's water-cooled flagship.

• The next generation bets on Samsung's 2nm process, not yet delivered — in a market where "falling behind on efficiency means you can't sell," persistent backwardness is a structural disadvantage.

Falling behind on efficiency = the customer (miner) pays more on electricity to run its machines, with a slower payback = it must cut prices to sell = thinner gross margin. This is the root of its single-digit or even negative gross margin.

5.2 Share: at the bottom (under any convention)

• Hashrate / installed-base convention (high-end figure): Bitmain about 80% > MicroBT about 15% > Canaan <5% (Cambridge University's figure about 2%).

• Broader shipment conventions (common at research firms): Bitmain 35%–45% > MicroBT about 25% > Canaan about 15%.

• Under any convention, Canaan is the bottom of the three giants. Its only "scarcity" lies not in the product, but in the fact that it is the only pure-play on this track that can be bought in the secondary market (Bitmain and MicroBT are both unlisted) — that is scarcity, but it is not a moat.

For reference: Bitmain's latest private-placement valuation is about USD 14.5 billion (PitchBook figure), about 54x Canaan's market cap (about 267 million) — the primary market's pricing of the industry leader throws into relief the secondary market's extreme distrust of Canaan, the bottom dweller.

5.3 Pricing power: zero

Rigs are standardized hardware, prices swing violently with the coin price, and Canaan has no say over the end price: coin price up means volume and price both rise; coin price down means it is forced to cut prices to clear inventory. A single-digit or even negative gross margin (Q1'26 gross margin was negative) is direct evidence of the absence of pricing power.

5.4 Composite: 2/5

Compared with names given "Watch": Laopu (brand moat 3), ILMN (installed-base lock-in / switching cost 4, a genuine chokehold position but eroding) — these are "real moat but expensive price / share being nibbled away." Canaan is a moat that is itself narrow, and continuously falling further behind rivals — a fundamental qualitative difference.

VI. Industry Demand: dominated by the single variable of Bitcoin price

6.1 The sole driver of demand: Bitcoin price

Rig demand is almost entirely determined by the price of Bitcoin, with the transmission chain being:

Coin price ↑ → mining profit ↑ → miners expand hashrate → rush to buy rigs → Canaan's volume and price both rise Coin price ↓ → mining profit ↓ / turns to loss → miners stop expanding or even shut down → rigs go unsold → Canaan's volume and price both fall

6.2 Current cycle position: bearish (most miners already mining at a loss)

• Bitcoin's current price is about USD 61,900 (2026-06-05), already halved by 51% from the October 2025 all-time high of about USD 126,000.

• After the 2024-04-20 halving, the block reward dropped to 3.125 coins — miners' unit output halved, structurally raising the mining cost.

• Listed miners' average cash cost is about USD 80,000–90,000 per coin — meaning at the current price of USD 61,900, most miners are mining at a loss.

This is the worst combination for rig demand: the coin price falls below miners' cost line → miners have no appetite to expand and may even shut down to cut output → demand for new rigs freezes. Canaan's Q2'26 revenue guidance of just 35–45 million (continuing to decline) is the direct embodiment of frozen demand.

6.3 Long term: the structural headwind of halving

Bitcoin's mechanism of halving every four years makes miners' unit revenue decline over the long run, offset only by "a rising coin price" or "improving efficiency." This is especially unfavorable to efficiency-lagging Canaan — it is on a track of long-term headwinds, still holding a shovel that is not sharp enough.

VII. Horizontal Analysis: simultaneously the cheapest, the weakest, and the most dangerous

7.1 Valuation comparables (2026-06-05)

7.2 The core comparable: Bitdeer (BTDR) — "the same hand, a far better way of playing it"

Bitdeer and Canaan are the most direct mirror images: both do the three-in-one of "self-developed ASIC chips + self-mining + AI cloud." But the pricing the market assigns is worlds apart:

• Bitdeer's market cap of about USD 4.25 billion is about 16x Canaan's (about 267 million).

• For the same business-model combination, the market is willing to pay 16x the price for Bitdeer — the gap comes from execution, hashrate scale (65.5 EH), technology roadmap, and survival certainty.

This comparison says it best: Canaan's cheapness is not because "the market hasn't discovered this business" (Bitdeer proves the same business can earn a high valuation), but because "the market has concluded that this company, Canaan, runs this business badly, and faces survival risk."

7.3 The paradox of "the largest revenue, the smallest market cap"

Canaan's TTM revenue (about USD 510 million) is larger than that of every one of WULF, IREN, CORZ, and APLD in the table above (their revenue is smaller in scale), yet its market cap is only 1/45 to 1/73 of theirs.

The only reasonable explanation for this paradox: what the market assigns is not a "revenue multiple," but a "business model + profit quality + survival certainty" multiple. AI-transition miners, though small in revenue now, are believed by the market to have sustainable high-margin, high-value AI-compute leasing; whereas Canaan's revenue, though large, is low-quality revenue that makes no money, is still loss-making, and faces delisting. Large revenue with a small market cap is precisely the quantitative embodiment of "the market pricing its operating business at zero."

7.4 Horizontal conclusion

Within the large family of rigs / miners / AI-transition names, Canaan simultaneously occupies three "mosts": the cheapest valuation (P/S 0.52, P/B 0.70), the weakest fundamentals (bottom share, lagging efficiency, consecutive losses), and the most dangerous survival (delisting clock + death spiral). Cheapness is the result of "weakest + most dangerous," not an independent investment opportunity.

VIII. Current Fundamentals

• Share price: USD 0.3581 (2026-06-05, −7.94%), a penny stock, near the 52-week low of 0.358 and the all-time low; 52-week range about 0.358–2.22.

• Market cap: about USD 267 million (data vendor's 6/5 figure of about 744.84 million ADS; the primary Q1'26 disclosure showed about 690.6 million as of 3/31, the difference being subsequent ATM issuance, with market cap floating along with issuance).

• EV: about USD 278 million.

• Valuation multiples: P/S (TTM) about 0.52x, P/B about 0.70x (already below net asset value per share).

• Cash / debt: cash about 43.5 million, interest-bearing debt about 54.5 million, no convertible bonds; equity about USD 382 million (still positive).

• Coin-hoarding: headline 1,807.60 BTC (only about 802.6 self-owned) + 3,951.53 ETH.

• Bitcoin: about USD 61,900, −51% from the October 2025 high.

• Business: 100% pure Bitcoin beta (the AI line was exited in June 2025).

• Delisting status: Nasdaq's second deficiency notice, with the grace period running to 2026-07-13.

8.1 Ownership structure and founder control

• Founder Nangeng Zhang controls the company through Class B super-voting shares (15 votes per share, held via the Flueqel Ltd. trust). His voting power fell from 72.6% at IPO to 46.2% in FY2024 (economic interest about 9%); it has fallen further with continuous dilution thereafter (estimated about 31%, not an official company disclosure, for reference only).

• Dilution is stark: Class A ordinary shares expanded from about 2 billion shares at IPO to about 11.2 billion shares (about 5.6x); ADS doubled from about 344 million to about 745 million within 18 months. ATM issuance (cap USD 270 million) is the continuous source of dilution.

• Judgment: the founder's economic interest is only about 9% and skin-in-the-game is on the low side, yet he firmly controls the company through super-voting shares — in a structure that continuously dilutes outside shareholders while being steered by a founder with little economic interest, protection of minority shareholders' interests is on the weak side.

IX. Valuation: cheap is a trap, and the current price is still expensive

9.1 Why DCF / P/E cannot be used

Canaan has consecutive losses, no positive free cash flow, and violently swinging revenue, so neither DCF nor P/E applies. The reasonable valuation anchors are asset residual value + relative valuation + scenario analysis.

9.2 Asset residual value (a liquidation lens)

• Equity is about USD 382 million (still positive), but a substantial part of it is rig inventory (highly cyclical, potentially writing down sharply when the coin price falls; Q1'26 already booked a 24.5 million inventory write-down), receivables (including the 905 BTC already pledged), and PP&E (mining-site equipment).

• P/B 0.70x means the market is willing to pay only 70 cents on the dollar for its book net assets — this itself is a signal that the market judges "the quality of book assets is questionable, and liquidation value is below book."

• The truly "clean" liquidation value (cash 43.5 million + unencumbered self-owned coins of about 802.6 ≈ USD 50 million − liabilities 54.5 million) is far below the book equity value — the "cheapness" of P/B 0.70 is highly dependent on whether the rig inventory can be realized at book, and that is extremely uncertain in a bear market.

9.3 Relative valuation

• P/S 0.52x is the lowest among peers, but for a loss-making, bottom-share name facing delisting, a low P/S is the "discount it deserves" rather than a "mispriced bargain" — the market rationally prices its operating business at zero (see the revenue-to-market-cap paradox in Section VII).

9.4 Scenario analysis

The current price of 0.3581 falls within the base range (toward the middle), meaning the market's current pricing already implies a neutral expectation of "coin price churns + barely surviving listing" — the current price leaves no discount for the tail risk of "death spiral / delisting failure," nor a cheap entry point reserved for a "major coin-price bull." In other words: the current price is not expensive enough for bears to short, and not cheap enough for bulls to have a margin of safety — an awkward position that "pleases neither side."

9.5 Sell-side targets (for reference only, with extreme divergence)

Several sell-side houses give "strong buy," with a target range of about USD 1.50–8.00 (implying 200%+ upside), but these are penny-stock sell-side targets of limited reference value: coverage is thin, varying widely with the data source / timing (e.g., one anchored figure of 1.69, MarketBeat about 3.21, Rosenblatt about 2.25). This report does not adopt sell-side targets as a valuation basis — for a penny stock facing delisting, "strong buy" reflects an industry-beta option more than the company's quality.

9.6 Valuation conclusion

Cheap is a value trap, and the current price is still expensive. The P/B 0.70 discount is highly dependent on realizing the rig inventory, and that assumption is fragile in a bear market; P/S 0.52 is a deserved discount, not a mispriced bargain. At USD 0.3581 there is no margin of safety. A symbolic ideal buy of ≤ USD 0.20 — it has speculative (not investment) value only when it falls to near the asset liquidation residual value and the two signals of "listing-survival confirmation + the start of a coin-price bull market" appear simultaneously.

X. Risks (including a Pre-mortem)

10.1 Core risk list

• Delisting risk (most pressing): Nasdaq's second deficiency notice, grace period 2026-07-13; the current price of 0.3581 needs +180% to return to USD 1, and having never done a reverse split historically, it will most likely be forced into a reverse split to save itself — and the reverse split itself is often another downside catalyst (retail selling, declining liquidity).

• Death-spiral risk: Coin price ↓ (already −51% from the high) × miners mining at a loss (cost 80,000–90,000 > current price 61,900) × frozen rig demand × continuous dilution, all interlocked and mutually reinforcing.

• Bitcoin-price risk (systematic): 100% pure Bitcoin beta; the coin price is the single most important variable, and it is unpredictable.

• Dilution risk: The ATM cap of USD 270 million draws down at any time, Class A shares have already expanded 5.6x, and every round of life support dilutes existing shareholders.

• Risk of persistent technological backwardness: Efficiency about one generation behind Bitmain / water-cooling nearly two generations, with the next generation betting on Samsung's 2nm not yet delivered; falling further behind means further share loss.

• Coin-pledge liquidation risk: 905 BTC are already pledged for collateralized loans, and a falling coin price could trigger margin calls and passive liquidation, amplifying the downside.

• Governance risk: The founder controls the company with super-voting shares despite an economic interest of about 9%, with weak protection of minority shareholders.

10.2 Risks that have been removed / are overstated (a reality check)

• HFCAA / China-ADR delisting risk has been removed: The PCAOB revoked on 2022-12-15 its inspection obstacles regarding China audit working papers, Canaan has no VIE structure, and its auditor KPMG Huazhen can be inspected — the market's concern about "China-ADR forced delisting" does not apply to Canaan and should not be counted as a current risk. The delisting risk in this case is the Nasdaq USD 1 minimum-price rule, which is an entirely different matter from HFCAA.

10.3 Pre-mortem: if this is a disaster in three years, the most likely cause

The most likely script: Bitcoin enters or stays in a bear market → miners broadly lose money and stop expanding → Canaan's rigs go unsold, self-mining loses money, and coins are marked down, a triple hit → cash depletes faster → forced high-frequency ATM issuance, with violent shareholder dilution → the July 2026 delisting deadline forces a reverse split → fundamentals unchanged after the reverse split, continuing to grind down → pledged coins trigger liquidation → trending toward shell value or even exiting via delisting. Every link in this script already has real-world signs, and that is the fundamental reason for "Avoid."

The reverse (bull) script: Bitcoin restarts a supercycle bull market → miners rush to buy rigs → Canaan's volume and price both rise, its coins appreciate sharply, and listing survival succeeds → the share price rebounds several-fold. This script is entirely possible (it is BTC high beta), but it is a lottery ticket, not a demonstrable investment — and that is precisely the essence of "a speculative chip" rather than "an investable security."

XI. Catalyst Tracking

11.1 Downside catalysts (warranting more vigilance)

• The 2026-07-13 delisting deadline: if still below USD 1 → forced reverse split / delisting process advances (a high-certainty event).

• Bitcoin breaking key support (such as USD 50,000) → miners' cost line further breached → freeze in rig demand intensifies.

• Cash falling below the safety line → high-frequency ATM issuance → violent dilution.

• Quarterly results continuing in gross loss / widening operating loss (Q2 guidance has already shown weakness).

11.2 Upside catalysts (lottery-like)

• Bitcoin restarting a bull market (breaking the prior high of 126,000) → rig supercycle + coin appreciation.

• A successful reverse split saving the listing + a cooperating coin price → valuation repair.

• The next-generation rig (Samsung 2nm) matches / overtakes on efficiency → share stabilizes (low probability of delivery).

11.3 Metrics to track

Bitcoin price (most important), Nasdaq compliance progress (2026-07-13), quarterly cash balance and rate of bleeding, the pace of ATM issuance, the rig-efficiency roadmap (Samsung 2nm), and self-mining hashrate and unit cost.

XII. Where Vertical and Horizontal Meet

Viewed vertically (history and share-price record): a value-destruction machine that has evaporated 96%–99% in six years, never reverse-split, and lost money for four consecutive years totaling about USD 963 million. The revenue roller coaster of +1041% then −67% with the coin price proves it sits on the cycle rather than riding it.

Viewed horizontally (peer comparison): within the large family of rigs / miners / AI-transition names, it simultaneously occupies three "mosts" — the cheapest (P/S 0.52, P/B 0.70), the weakest (bottom share, lagging efficiency), and the most dangerous (delisting + death spiral). The mirror-image name Bitdeer has a market cap 16x its own, and it has the largest revenue yet the smallest market cap, quantitatively proving that "the market prices its operating business at zero."

The convergent conclusion: The vertical "history of value destruction + highly cyclical, no pricing power" and the horizontal "cheapest = least trusted" corroborate each other, jointly pointing to the same qualitative judgment — this is a bad business, layered with high speculation and a survival-level tail risk. With the now-bankrupt AI narrative stripped away, Canaan is a pure Bitcoin high-beta speculative instrument: its "cheapness" is the result of rational market pricing, not a mispricing; its upside is a lottery ticket dependent on a major coin-price bull, not a demonstrable investment return.

Rating: Avoid. Different from both "Watch" (good business + expensive price) and "Sell" (collapsing fundamentals with no residual value / overvalued): this case is a bad business + high speculation + survival risk, and rational investors should stay away; but because equity is positive, debt is light, HFCAA is removed, most bad news is priced in, going to zero is not the base case, and it retains residual value as a deep-cyclical BTC option, it does not reach "Sell." It is not an investable security; it is a speculative chip. A symbolic ideal buy of ≤ USD 0.20; at USD 0.3581 there is no margin of safety whatsoever.

Research Uncertainties

• Market-cap / share-count convention: This report's market cap of about USD 267 million is based on the data vendor stockanalysis's 2026-06-05 figure (about 744.84 million ADS); the company's primary Q1'26 filing disclosed about 690.6 million ADS as of 2026-03-31, with the difference being mainly subsequent ATM issuance. Market cap floats continuously with issuance; the precise figure follows the latest 6-K / 20-F.

• The founder's current voting power: 72.6% (IPO) → 46.2% (FY2024) is company-disclosed; the subsequent "about 31%" is an estimate based on continuous dilution, not an official company disclosure, for reference only.

• Divergence in market-share conventions: The hashrate / installed-base convention (Bitmain 80% / MicroBT 15% / Canaan <5%) differs substantially from broader shipment conventions (35%–45% / 25% / 15%); this report flags each instance and does not average them; Cambridge University's hashrate figure gives Canaan about 2%.

• The two cumulative-loss conventions: The "sum of four consecutive annual losses" of about 963 million differs from the balance-sheet "accumulated deficit" of 749.5 million; they have been distinguished throughout and must not be conflated.

• Bitcoin price: About USD 61,900, −51% from the high, is a 2026-06-05 multi-source cross-referenced figure; the coin price is highly volatile and may have changed by the time of reading.

• Coin-hoarding breakdown: The self-owned / receivable / pledged split of the headline 1,807.60 BTC is based on company disclosure and multi-source cross-referencing; the specific terms (margin-call lines, etc.) of the pledges and fixed-term products are sparsely disclosed.

• Sell-side targets: Coverage is thin, varying widely with the data source / timing (the USD 1.50–8.00 range); this report does not adopt them as a valuation basis, only as an industry-beta sentiment reference.

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