Cipher Digital: A Deep Value Investment Study

Hunting ten-year five-baggers among great growth stocks — pressing the upside question: "Can it get much bigger?"

• QHow high is its market ceiling? Is it growing an existing pie, or creating an entirely new market?5/10

  A

  The ceiling is very high, but Cipher is "growing a slice of an already-existing, crowded pie" (AI/HPC data center capacity), not creating an entirely new market — and the share it can actually claim today is still just a promise on paper, not built-out capacity.

  Looking at the industry ceiling, the space itself has real and enormous room to grow. The report cites McKinsey's estimate: global data center construction will require about $6.7 trillion in capex through 2030, of which the AI-related portion is about $5.2 trillion. I independently verified this figure — McKinsey's original text does indeed say data centers will need about $6.7 trillion in capex through 2030, of which about $5.2 trillion is specifically for AI compute, corresponding to about 156GW of AI data center capacity demand. The supply side is equally tight: CBRE data cited in the report shows data center vacancy in major North American markets falling to a historic low of 1.4% in H2 2025, and the IEA expects global data center electricity consumption to rise to about 945TWh by 2030. So on the question of "is the industry ceiling high," the answer is a clear yes.

  But what Baillie Gifford is really asking is "can this company create a brand-new market of its own" — and on that count, Cipher doesn't qualify. What it does is lease power, land, interconnection, and construction capability it has already locked up to hyperscalers like Amazon and Fluidstack/Google — essentially fighting for a slice of the existing "data center colocation/build-to-suit" pie, competing against IREN, Applied Digital, traditional REIT-style data center operators, and hyperscalers building their own capacity. Unlike the names Baillie Gifford favors, it isn't defining a brand-new category or reshaping demand structure — it's a capacity supplier: how many MW a customer wants, at what price, and for how many years is largely up to the hyperscaler.

  More importantly, the "slice of the pie" Cipher can currently claim still sits at the contract stage. I verified the figures disclosed in the company's Q1 2026 business materials: 700MW of contracted HPC capacity, about $11.4 billion in contracted revenue, and about $787 million in average annualized NOI (made up of three hyperscaler campus leases — Barber Lake, Black Pearl, and one more — with base terms of 10-15 years). That set of numbers is itself the upper-bound anchor for its "share of the pie"; but the reality the report keeps emphasizing is that, as of Q1 2026, the company's financials still show only one operating segment, Bitcoin Mining, with Q1 revenue of $34.84 million entirely from mining. In other words, the ceiling is written into the leases, while the floor still sits on mining rigs — with a whole bridge of construction, delivery, rent commencement, and financing realization in between.

  My judgment: this is a story of "grabbing a slice of an existing, huge, but extremely capital-intensive pie with many players, using a power/land head start" — not the kind of disruptive, market-creating growth Baillie Gifford is looking for. The pie is big enough to make the upside case real, but Cipher neither defined this pie nor has yet proven it can reliably hold onto its own slice — its ceiling height is not the same thing as its certainty of realization.

  Jun 10, 2026
• QCan its revenue at least double over the next five years? Is growth mainly driven by volume, price, or new business?4/10

  A

  Probably yes — and by far more than double — but this "doubling" is almost entirely an accounting step-change driven by new business (HPC rent) ramping up from zero, not the existing mining business growing organically through volume and price. So its certainty hinges on engineering delivery, not market demand.

  First, look at the gap between the base and the target. The report discloses 2025 revenue of $223.9 million, while Q1 2026 revenue has already shrunk to $34.84 million (annualizing to about $140 million), because Black Pearl's mining operation was shut down as part of the conversion. I verified this step-down: Q1 2026 revenue of about $35 million, down from about $60 million in Q4 2025, due to Black Pearl mining winding down and the transition toward contracted data-center revenue. In other words, looking only at the existing mining segment, revenue isn't doubling — it's deliberately shrinking.

  What will actually push revenue up is the "rent commencement" of the signed HPC leases. The report and company materials agree: 700MW of contracted capacity corresponds to about $11.4 billion in contracted revenue and about $787 million in average annualized NOI (commencing October 2026 through September 2036). Once campuses such as Barber Lake and Black Pearl are delivered and reach rent commencement, an annualized NOI figure of about $787 million is several multiples of today's $100-200 million revenue base — making "at least doubling" within five years an almost low bar mathematically, provided the projects are built on time.

  So the growth breakdown is clear: it's not volume (mining more coins), not price (Bitcoin rising), but new business (HPC rent) converting from contract to cash flow. That is precisely what distinguishes it from a "natural growth stock" — growth doesn't come from demand elasticity or accumulating pricing power, but from a one-off asset-delivery event. The report spells out the fragile point very clearly: management lists "failure to complete HPC data center construction on time or on budget" as a core risk in its 10-Q, and tenant guarantees and backstops only truly take effect after rent commencement. I verified the leases are real and their scale is accurate — both the roughly $5.5 billion, 15-year, 300MW lease with Amazon (AWS) and the Fluidstack (Google backstop) Barber Lake lease at a full 300MW, totaling about $3.8 billion genuinely exist — not slideware.

  One honest hedge to add: because the revenue step-change depends on engineering milestones rather than the market, it carries both an "overshoot" tail and a "badly miss" tail — campus delays, lease renegotiation, or mining revenue collapsing faster than new rent ramps up during the transition could all cause revenue to fall in some quarters rather than rise (Q1 is exactly an example). But if the question is "can revenue double on a five-year horizon," as long as most of the 700MW lands broadly on schedule, the answer is yes; if the question is "is this doubling high-quality organic growth," the answer is no — it is a capitalized revenue leap driven by contracts, realized through construction, and supported by financing.

  Jun 10, 2026
• QFive years from now, what will take the baton as the next growth engine? Does this "second curve" exist today?4/10

  A

  A bit awkwardly: for Cipher, what the market perceives as its "second curve" (HPC data centers) is actually its first curve that it hasn't fully climbed yet — the real next growth engine is the further-out site-pipeline realization, and today that curve exists only in the form of a "power/land reserve," far from signed and even further from built.

  To untangle this question, we first need to correct a sequencing mismatch. Baillie Gifford's question of "does a second curve exist today" usually assumes the first curve is already generating cash in a steady state. But Cipher's peculiarity is that its only current cash-generating "first curve" is Bitcoin mining, which is being actively wound down (Q1 2026 revenue of only $34.84 million, and declining as Black Pearl is retired), while the HPC rent the capital markets treat as the main engine hasn't contributed a single dollar of revenue in the financials as of Q1 2026. So strictly speaking, Cipher is currently in a gear-shift vacuum where "the first curve is receding and the second curve (HPC) has not yet commenced" — talking about a "third curve" is somewhat premature.

  So, assuming this HPC curve successfully becomes the core business within five years, what takes the baton after that? The report's answer is further signing and monetization of the pipeline reserve. The company's annual report repeatedly emphasizes a roughly 4.2GW site reserve, and investor materials show a current pipeline of about 3.3GW to 4.2GW spread across Texas, Ohio, and other locations, covering projects such as Barber Lake, Black Pearl, Ulysses, Colchis, Milsing, and McLennan. In other words, beyond the signed 700MW there is several times that much in "power + land" optionality — that, in theory, is the next growth engine.

  But the extent to which this curve "exists today" is very shallow, and that must be honestly flagged. Its current form is an unsigned capacity reserve, not locked-in contracted revenue: the 700MW is the signed portion behind about $11.4 billion in contracted revenue, while the remainder of the 3.3-4.2GW after subtracting that 700MW has no customer, no lease terms, and no NOI anchor. The report's bullish scenario treats "at least one more major project in the pipeline being successfully signed and monetized" as a precondition for intrinsic value reaching $24-30 — which is precisely why this piece of value is "optionality," not "in hand."

  The more critical sustainability question: whether this curve can keep passing the baton depends on two capabilities Cipher has not yet proven. The first is financing regenerability — every additional GW-scale campus it signs requires another round of project debt or equity dilution (the report already notes about $3.19 billion in net financing inflows in 2025 and another roughly $1.96 billion added in Q1 2026, plus significant potential dilution from convertible notes and Google warrants) — growth is being bought with the balance sheet, not compounded through self-generated cash. The second is the replicability of power access — SB 6, passed by Texas in 2025, will impose new process requirements, security deposits, and interconnection conditions on large loads above 75MW, and ERCOT batching may also slow the pace, meaning that "securing the next cheap block of power" is itself getting harder.

  My judgment: Cipher does not have a second curve in the Baillie Gifford sense — one that is "already incubating and independently observable." What it has is "continual replication of the same business model (power arbitrage → lease to hyperscalers)." The real next engine is pipeline signings, which today exist only in the potential form of land and power reserves, with every extension needing to be ignited by new financing — this is a growth path that requires continuous external transfusion and has not yet proven it can self-perpetuate, rather than a new curve that has grown organically.

  Jun 10, 2026
• QWhat is its core competitive advantage? Will this moat widen or narrow over the next three to five years?4/10

  A

  The core advantage is a "power-first" first-mover position — securing cheap power, land, interconnection, and construction/organizational capability ahead of everyone else, at a time when the whole industry is bottlenecked on power interconnection and land. This is a real and scarce advantage that could theoretically widen over the next three to five years (if campuses are delivered and leases renew reliably), but today it does not yet amount to a "durable moat" — it is closer to an execution window that will narrow over time.

  First, what exactly is this advantage, and how solid is it. The report points out that Cipher's most realistic barrier is not brand or network effects but power-first site-selection capability: the Odessa site's fixed power price after October 2025 is about 2.8 cents/kWh, and the company also holds a roughly 4.2GW site reserve. I independently verified the power price — the company's Q1 2026 earnings call confirmed Odessa's power cost at about $0.028/kWh, which, given today's extremely tight power environment, is a genuine cost advantage. The report cites CBRE data showing North American data center vacancy falling to a historic low of 1.4% in H2 2025, meaning that "whoever holds an already-energized, interconnect-ready campus" is itself a scarce resource — this is the source of Cipher's first-mover value.

  But breaking the moat down across the five classic dimensions, Cipher scores weak on most of them, and this must be stated honestly:

  • Brand: weak. Customers are hyperscalers like Amazon and Fluidstack/Google — competition is about sites, power, and delivery speed, not brand recognition.
  • Network effects: almost none. Each campus is an independent physical asset that doesn't become more valuable as more users join.
  • Switching costs: there is some physical relocation cost for tenants already signed to long-term campus leases, but this is a general characteristic of data center assets, not unique to Cipher.
  • Patents/licenses/data: the report explicitly notes the annual report offers no evidence sufficient to constitute a core barrier.
  • Cost and scale advantage: this is the one dimension with real substance — a 2.8-cent power price plus in-house development/engineering/procurement/operations capability does genuinely speed up delivery, but "cheap power" is not permanently exclusive, and new entrants are also competing for power in Texas and Ohio.

  Will it widen or narrow over the next three to five years? The two business lines point in opposite directions. For the legacy mining business, the moat is narrowing — the report discloses that the company revised its expected rig useful life from five years to three years in mid-2024, and in 2025 took $45.3 million in equipment impairment and a $96.1 million write-down on held-for-sale assets, indicating that both the economic life and use-value of mining rigs are depreciating. I verified the structural drag from the halving on unit economics: the industry backdrop of the 2024 Bitcoin halving cutting the block subsidy from 6.25 BTC to 3.125 BTC further erodes miners' revenue per unit of hashrate. For the new HPC business, the moat could theoretically widen — once campuses are delivered, leases (15-year terms) commence, and customers renew reliably, a single-tenant long-term lease creates a fairly solid form of lock-in. I agree with the report's judgment here: this is a "moat still to be proven," not an established one.

  There's also a structural factor that discounts the moat: the power-access head start is getting harder to secure, casting doubt on the moat's "renewability." The report notes that SB 6, passed by Texas in 2025, will impose new process requirements, security deposits, and interconnection conditions on large loads above 75MW, and ERCOT's large-load batching will also affect the pipeline's pace. In other words, today's power first-mover advantage comes partly from the time window of "entering early," and that window is tightening due to regulation and competition.

  My judgment: Cipher's moat is of the "asset first-mover + execution window" type — real and valuable today, but its source is timing and capital, not a structural, non-replicable barrier. Its trajectory over three to five years forks two ways: if the 700MW is delivered smoothly and renews reliably, the HPC-side moat will widen substantially; if delivery is delayed or power access is blocked, it looks more like a cyclical developer whose edge will narrow as the window closes. Betting on it today essentially means betting that it can pour and set its lease moat before the window shuts.

  Jun 10, 2026
• QIf its core business were disrupted, does it have a gene for self-reinvention? How does it handle mistakes and bad news?5/10

  A

  It has a rare gene for self-reinvention that has already been proven once — the "Bitcoin miner → AI data center landlord" transformation Cipher is currently undergoing is itself a case of proactively disrupting its own core business; its way of handling bad news leans toward "decisively cutting losses, taking write-downs, and pivoting" rather than concealment. But the underlying nature of this gene is high-risk capital engineering, not the "culture-driven antifragility" Baillie Gifford favors, and the capability it depends on most — continued financing — is also its single biggest source of fragility.

  Let's start with this implicit premise of a "self-reinvention gene" — this is Cipher's most standout feature. Most companies are unwilling to disarm themselves while their first curve is still making money; Cipher, by contrast, proactively shifted its core sites from mining to HPC while Bitcoin mining was still its only source of cash. The most compelling evidence is that it shut down assets that were actively producing coins to free them up for leases — I verified this: Black Pearl's mining operation was fully retired in February, causing Q1 2026 company revenue to fall from about $60 million to about $35 million sequentially, with management explicitly stating no further capital would be invested in the mining business. Being willing to cut cash flow for a second curve while the first curve is still alive is a genuine turnaround resolve, not just talk. The report also confirms the execution: signing 300MW/15 years with Amazon at Black Pearl, a long-term lease with Fluidstack (Google backstop) at Barber Lake, and a third investment-grade hyperscaler lease in Q1 — showing this "secure land → secure power → sublease" reinvention capability is repeatable.

  But an honest distinction is needed: this is "capital-allocation flexibility," not the kind of gene Baillie Gifford means by "growing a new species out of its culture after being disrupted." Cipher's reinvention rests on CEO Tyler Page's finance background (a résumé spanning NYDIG, Stone Ridge, Guggenheim, and Goldman Sachs) and aggressive project-financing capability, not an engineering/product culture that can keep self-iterating through dramatic technology shifts. Its "regeneration" is highly dependent on capital markets remaining willing to keep feeding it — the report discloses about $3.19 billion in net financing inflows in 2025, with another roughly $1.96 billion added in Q1 2026. Once the financing window closes (AI capex cooling, high interest rates), this externally fueled reinvention machine would stall. In other words, its self-reinvention capability is "strong in up-cycles, weak in down-cycles" — the exact opposite of the antifragility Baillie Gifford wants.

  Next, look at "how it handles mistakes and bad news" — here Cipher performs relatively honestly and doesn't dodge. There's several pieces of evidence. First, it admits rig depreciation — revising rig life from five years down to three in mid-2024, and taking $45.3 million in equipment impairment and a $96.1 million held-for-sale write-down in 2025, effectively an open admission that heavy-asset use-conversion is more fragile than expected. Second, risk disclosure isn't sugarcoated — the report notes the company proactively lists "failure to complete HPC construction on time/on budget" and "single-tenant properties, where losing either Amazon or Fluidstack/Google could cause a material adverse effect" as core risks in its 10-Q, without using the AI narrative to paper over concentration and delivery risk. Third, the accounting isn't dressed up — the report shows that much of the 2025 net loss of $822 million consists of non-cash items like impairments and fair-value changes, which the company discloses faithfully rather than glossing over. This attitude of "putting bad news on the table" is more honest than many peers who are propping up valuations with slide decks during the same period.

  One hedge, though: candid risk disclosure and "governance that genuinely puts shareholders first" are two different things. The report notes the CEO's 2025 total compensation was about $14.969 million (of which stock awards were about $12.480 million), the say-on-pay approval vote was only about 72%, and insiders' actual shareholding alignment isn't especially heavy (CEO ownership about 2.10%). Against a backdrop of continued large losses and shareholder dilution, this compensation arrangement suggests management isn't especially restrained about being generous to itself — which erodes trust that shareholder interests will be prioritized when bad news hits.

  My judgment: Cipher has real, once-proven self-reinvention capability (proactively disrupting its own mining core business), and it also leans toward decisive handling and honest disclosure of mistakes and bad news — this is a plus, and a key thing that distinguishes it from pure-play miners. But the underlying gene rests on "continuous financing plus aggressive capital engineering," fragile in down-cycles and not restrained enough toward shareholders — it can pivot nimbly in a tailwind but may not necessarily protect long-term shareholders in a true make-or-break crisis.

  Jun 10, 2026
• QDoes management (especially the founder) have a long-term view, and are their interests deeply aligned with the company? Are they willing to sacrifice near-term profit for the sake of five-to-ten years out?5/10

  A

  There is a long-term view, and a willingness to sacrifice near-term profit for the sake of five to ten years out — Cipher has been quite thorough about this. But the criterion of "interests deeply aligned with the company" does not pass — the founder-CEO's actual shareholding is on the low side, compensation is not light, and shareholders are being continuously diluted; together these three factors discount the alignment of interests.

  Let's start with the highlight: on whether it's willing to sacrifice the present for the long term, Cipher is a rare name among its peers that "dares to look bad right now." The whole company's current free cash flow is deeply negative and net losses are enormous, and the root cause is precisely that management has put all resources into 5-15-year HPC campus construction, proactively shutting down coin-producing mining assets. I verified how forceful this trade-off is: the company explicitly states it will make no further capital investment in the mining business, with Black Pearl mining fully retired in February, causing revenue to fall sequentially from about $60 million to about $35 million. A company willing to let its current-period financials look this bad for the sake of a future lease shows it is genuinely allocating resources on a 5-10-year scale rather than managing to quarterly EPS — which is exactly the long-term view Baillie Gifford appreciates. The 700MW, roughly $11.4 billion in contracted revenue, and roughly $787 million in annualized NOI disclosed in the report will all only begin commencing after October 2026 — that is itself clear proof of "sacrificing today for cash flow a decade out."

  On whether the founder is still at the helm and holds a long-term view, the answer also leans positive. The report notes CEO Tyler Page is the company's founder, with a résumé spanning NYDIG, Stone Ridge, Guggenheim, and Goldman Sachs; management and the board as a whole come from backgrounds in traditional finance, digital assets, trading, and data centers. The founder is still steering the ship and has personally led this extremely high-risk strategic transformation — a positive signal for a "long-termist at the helm."

  But on the core gate of "interests deeply aligned with the company," Cipher does not pass, and that must be stated honestly. The shareholding data in the report is telling: as of April 8, 2026, the CEO held about 8.507 million shares, or 2.10%, and all directors and executive officers together held about 12.529 million shares, or 3.09%. Against the scale of the company's market cap — I verified the current market cap has already reached about $9.4 billion (CIFR around $23, about 409 million shares, as of 2026-06-09) — the founder's 2.10% stake, while not negligible, doesn't carry much weight relative to this company's extremely high construction-period risk and enormous market cap in terms of insiders having "their own money on the table." Baillie Gifford favors founders who stake their fortunes and share fate with minority shareholders; Cipher's degree of alignment can only be described as "present but not deep."

  Going further, "deep alignment" is diluted by two more things. First, compensation is on the heavy side: the report discloses the CEO's 2025 total compensation was about $14.969 million (of which stock awards were about $12.480 million), and the say-on-pay approval vote was only about 72% — not an overwhelming margin. For a company with continuous large losses and deeply negative free cash flow, this level of compensation shows management isn't stingy toward itself. Second, shareholders keep being diluted: the report notes 2025 ATM issuance sold about 33.3 million shares at an average net issue price of only about $5.88/share, and together with the two convertible-note tranches (2030 conversion price about $4.45, 2031 about $16.03) and Google warrants covering 24.179 million shares, the two convertible tranches plus warrants alone correspond to roughly 144 million potential new shares — about a 35% increase relative to roughly 405 million shares outstanding. When management draws heavy, mostly stock-based pay while continuously diluting outside shareholders, the quality of "aligned interests" is weakened — their wealth comes more from granted equity than from buying in with real money on the secondary market and bearing dilution the way you do.

  My judgment: Cipher's management scores highly on "long-term view" and "willingness to sacrifice near-term profit for the long term" — they are genuinely playing a 5-10-year game, and the founder is still at the helm. But on the criterion Baillie Gifford cares about most — "interests deeply aligned with the company" — it fails: the founder's low shareholding, heavy compensation, and continuous dilution of outside shareholders together significantly discount the alignment of interests between management and minority shareholders. You can trust their capability and resolve, but you can't yet say their wallet is tied to the same boat as yours.

  Jun 10, 2026
• QIf it disappeared tomorrow, how much would customers miss it? Is its growth model sustainable, without relying on harm to society or exploiting regulation?4/10

  A

  Looking at the two dimensions separately gives opposite conclusions. Indispensability: weak — if Cipher disappeared tomorrow, its hyperscaler customers would feel short-term pain but wouldn't truly miss it, because what Cipher provides is replaceable capacity, not an irreplaceable capability. Social and regulatory sustainability: neutral-to-positive — the new HPC data center path is essentially harmless to society (indeed, it's essential infrastructure for the AI era), but it is highly dependent on power and regulatory approvals, and carries real risk of being constrained by the grid/policy.

  First, "if it disappeared tomorrow, how much would customers miss it." This question is testing indispensability, and Cipher's customer structure is exactly what exposes its replaceability. The report notes the future HPC business is a single-tenant campus model, with customers highly concentrated among a handful of hyperscalers like Amazon and Fluidstack/Google. To these customers, Cipher is "one of many capacity suppliers" — they have a long list of options including Cipher, IREN, Applied Digital, and can also self-build. I verified the replaceable nature of the lease: the roughly $5.5 billion, 15-year, 300MW agreement with Amazon buys "turnkey space and power," not some technology or service unique to Cipher. If Cipher exited, customers would lose the delivery timeline of this batch of already-energized campuses (there is real switching cost and time loss), but they would lose nothing that only Cipher can provide. Compared with names Baillie Gifford favors — the kind of indispensability where "a customer's business grinds to a halt if the company disappears" — Cipher falls far short.

  More importantly, the direction of this indispensability is actually reversed: it's Cipher that needs these few customers more, not the customers that need Cipher. The report puts it bluntly: the company explicitly states its HPC data centers have to date been single-tenant properties, and losing either Amazon or Fluidstack/Google could have a material adverse effect on operations and cash flow. The bargaining scale tips toward the hyperscaler side — they are the "indispensable" party. So the honest answer to this question is: Cipher would miss its customers very much; the customers wouldn't necessarily miss Cipher.

  Next, on the implicit premise of "is the growth model sustainable, without relying on harm to society or exploiting regulation" — here Cipher actually holds up. Unlike some businesses that expand through regulatory arbitrage, harm to consumers, or skirting policy red lines, Cipher's HPC transition is fundamentally physical infrastructure providing the compute base for AI — its social utility is clear, and there's no negative-externality problem of "the faster it grows, the more outside harm it causes." It is cutting away from mining's old image of "consuming power for crypto, with contested social value" and shifting toward data-center infrastructure that is more broadly accepted — an improvement in social acceptability, not a deterioration.

  But the "regulatory sustainability" dimension carries genuine uncertainty that shouldn't be glossed over. Cipher's lifeline is power, and power happens to be the most heavily regulated link. The report specifically flags that SB 6, passed by Texas in 2025, will impose new process requirements, security deposits, and interconnection conditions on large loads above 75MW, potentially raising costs and delaying timelines; ERCOT's large-load interconnection batching and stability requirements could also affect the pipeline's pace. I verified the industry-level rigidity of the power constraint — the IEA expects global data-center power consumption to rise sharply by 2030, with grid strain being a widespread issue. In other words, Cipher's growth sustainability doesn't hinge on "whether it harms society" (largely it doesn't), but on "whether the grid and regulators will allow it to keep adding large-scale power draw." This is an exogenous constraint it cannot control itself — a real risk point for sustainability, but not a moral/social demerit.

  My judgment: Cipher clearly fails on "indispensability" — what it provides is replaceable capacity, bargaining power sits with customers, and if it disappeared tomorrow, customers would switch suppliers rather than truly miss it; but on "social and regulatory sustainability" it is neutral-to-positive — its growth model doesn't harm society and is essential AI infrastructure, with the only hard constraint being the exogenous variable of power-interconnection regulation and grid capacity. Overall, on this Baillie Gifford question, Cipher's weak point is "indispensability," not "sustainability."

  Jun 10, 2026
• QWhat are this business's unit economics (gross margin, incremental returns)? Do they improve or worsen as it scales? Where does the money it earns go?4/10

  A

  The unit economics of the two business lines are two extremes: mining is a bad business with low margins, poor incremental returns, and gets worse as rigs depreciate; the HPC leases are, in theory, a good business with high NOI margins and contractual lock-in. Unit economics should improve in theory as scale grows (provided the leases are realized), but the money it earns is currently almost entirely poured into heavy-asset construction, with not a cent flowing back to shareholders — this is the fundamental difference from the "asset-light, high-incremental-return" model Baillie Gifford favors.

  First, the mining business's unit economics — this part is poor and worsening. The report discloses a 49.2% gross margin in Q1 2026, which looks decent, but mining's incremental returns are bad: every additional unit of hashrate carries the drag of rig depreciation, power costs, and rising network difficulty. I verified the structural facts dragging down unit economics — the 2024 Bitcoin halving cut the block subsidy from 6.25 BTC to 3.125 BTC, structurally cutting miners' output per unit of hashrate; combined with the report's disclosure that rig life was revised down from five years to three, and $45.3 million in equipment impairment was taken in 2025, mining's "incremental return on investment" keeps getting worse the more it scales. So the larger this business gets, the heavier the depreciation burden — a classic case of bad unit economics.

  Next, the HPC lease's unit economics — this part is good in theory, and it's exactly why the market is willing to pay a premium. The report and company materials agree: 700MW of contracted capacity corresponds to about $11.4 billion in contracted revenue and about $787 million in average annualized NOI, tied to 10-15-year long-term leases. Data center colocation NOI margins are inherently high, and I verified the industry benchmark — the company discloses site NOI margins of about 85%-90%, with project cost of about $9-10 million per MW. Once campuses are delivered and rent commences, this is contracted, predictable, high-margin cash flow, with unit economics far better than mining. So the answer to "does it get better or worse as it scales" is: the process of the business mix shifting from mining to HPC is itself the process of unit economics moving from bad to good — but that comes from the structural shift, not from the same business naturally improving as it scales.

  But there's a key flaw here that Baillie Gifford would immediately press on: the "denominator" of incremental returns is extremely heavy. An 85-90% NOI margin sounds attractive, but it rests on upfront capital investment of about $9-10 million per MW — 700MW corresponds to several billion dollars of construction capital. The report discloses capex as high as $487.9 million in 2025, and a single quarter's capex of $554.0 million in Q1 2026. In other words, this business must first swallow huge amounts of capital before it can earn the high-margin rent that follows; its incremental return on capital (ROIC) is deeply negative during construction, and could only turn positive years later once rent commences. This is the mirror image of the asset-light compounding machine Baillie Gifford loves — "invest one more dollar, get a high return immediately" — Cipher instead follows a capital-intensive model of "invest heavily first, collect late, amortize slowly through long-term leases."

  Finally, look at "where the money it earns goes" — this tells the story most clearly. The answer is: the money hasn't just failed to flow back to shareholders, it's actually "borrowed money plus issued money" being poured together into construction. The report discloses about $3.19 billion in net financing cash inflows in 2025, with another roughly $1.96 billion added in Q1 2026, going almost entirely to fixed-asset construction. I verified the debt scale — as of Q1 2026 total debt principal is about $5.2 billion, mostly non-recourse project debt, with about $3.5 billion of restricted cash earmarked for project construction. In other words, at this stage every dollar earned and raised is being absorbed by the capital black hole of construction, with no dividends, no buybacks, and deeply negative free cash flow. The report estimates conservative mature Owner Earnings of about $350-450 million per year, but that is a far-out picture "once projects are fully operational and excluding any additional pipeline" — today not a single dollar of it has landed yet.

  My judgment: Cipher's unit economics are a process of "a bad business (mining) being replaced by a good business (long-term HPC leases)" — the HPC side's roughly 85-90% NOI margin is genuinely excellent, and unit economics should improve in theory as it scales; but its incremental returns carry an extremely heavy capital denominator, ROIC is deeply negative during construction, and all the money it earns and borrows is poured into heavy assets with zero flowing back to shareholders. It isn't the asset-light, high-incremental-return compounding machine Baillie Gifford likes — it's a capital-intensive infrastructure model of "trading the balance sheet for long-term leases now, then slowly amortizing the return over a decade" — whether that's good or bad depends on whether those expensive campuses can ultimately deliver their high NOI into free cash flow as contracted.

  Jun 10, 2026
• QFor it to rise five-fold over ten years, what conditions need to hold simultaneously? Are these conditions realistic? What expectations does today's share price imply?3/10

  A

  For a ten-year five-fold rise, four or five high-difficulty conditions need to hold simultaneously, and none of them is fully within the company's own control; at the current share price (already up meaningfully from the report's benchmark), the market has already priced in a great deal of "successful transformation plus expansion far beyond the signed 700MW," leaving very thin cushion for a ten-year five-fold gain. Let's first pin down the price benchmark — the report anchors on about $18.80, market cap about $7.62 billion (2026-05-20), but I verified that CIFR has since risen to about $23.04, market cap about $9.42 billion, about 409 million shares (2026-06-09), up more than 20% over roughly the past month. That means the judgment made at the time of writing — "near the top of the base-case range, having already borrowed against the optimistic scenario" — is even more borrowed-against today.

  Conditions that need to hold simultaneously for a ten-year five-fold rise:

  First, the signed 700MW must be fully delivered on time, on budget, and on lease terms, and reach rent commencement. The report lists this as the number one uncertainty, and the company itself lists "failure to complete HPC construction on time/on budget" as a core risk in its 10-Q. I verified the signed capacity anchor genuinely exists — 700MW, about $11.4 billion in contracted revenue, about $787 million in average annualized NOI — but contracted revenue is not the same as cash received; there is several billion dollars of construction and multiple years of build-out in between.

  Second, NOI must truly flow through to distributable free cash flow, rather than being swallowed by interest and dilution. The report estimates conservative mature Owner Earnings for the signed projects at about $350-450 million per year (after deducting cash interest of about $220-250 million, corporate overhead, and maintenance capex). Note this net figure is far smaller than the $787 million NOI — leverage eats up roughly half the NOI.

  Third, the signed 700MW alone is not enough to support a five-fold rise — several more GW-scale major projects from the pipeline must be signed and realized. This is a hard mathematical constraint: a ten-year five-fold rise means market cap rising from about $9.4 billion to about $47 billion. Even applying an optimistic multiple of 15-20x Owner Earnings for a mature data-center platform, a $47 billion market cap would require about $2.4-3.1 billion per year of Owner Earnings — about 6-8x the $350-450 million corresponding to the signed 700MW. In other words, the vast majority of today's 3.3-4.2GW pipeline would need to be signed into investment-grade long-term leases and built out to support a five-fold gain. That is equivalent to requiring it to successfully replicate the entire transformation five or six more times.

  Fourth, all of this must happen under the premise of "no major additional dilution." This is nearly the hardest condition — the report has already noted that the two convertible-note tranches (2030 conversion price about $4.45, 2031 about $16.03, both deeply in the money at the current $23 share price) plus Google warrants covering 24.179 million shares alone correspond to roughly 144 million potential new shares — about a 35% increase relative to roughly 405 million shares outstanding. Every additional GW built means another round of project debt or issuance. Building out several GW while keeping per-share dilution under control is an extremely difficult capital-engineering feat.

  Fifth, the market's valuation framework must switch entirely from "miner/cyclical stock" to "premium data-center platform" and be awarded a high multiple. This is a narrative-level condition and the least controllable one.

  Are these conditions realistic? Taken individually, none of them is far-fetched — Cipher has indeed signed real leases with Amazon and Fluidstack/Google, proving its execution ability. But requiring all five to hold simultaneously, and to keep holding for ten straight years without a major slip-up, isn't very probable in reality. Any single link failing (campus delay, lease renegotiation, a round of low-priced issuance, AI capex cooling) breaks the five-fold path. This is a "multiplication of probabilities" — even at 70% success for each link, five links compounded leave only about 17%.

  What expectations does today's share price imply? At the current market cap of about $9.4 billion, against the report's estimated conservative mature Owner Earnings of $350-450 million per year, that implies about 21-27x mature, and still unrealized, Owner Earnings — the market has not only already priced the 700MW as fully built and fully rent-commenced as settled fact, it has additionally paid an option premium for "the pipeline continuing to be realized." The book-value picture is even more extreme: the report calculates P/B of about 10.7x at $18.80; recalculated at today's $23 against Q1 shareholders' equity of about $714 million, P/B is already about 13x. I verified the peer comparison — IREN's P/B is about 7.8x, MARA's is about 2.2x, and IREN has already disclosed >4.5GW secured power, 810MW operational, and 2.1GW under construction — its "degree of proof" is markedly stronger than Cipher's. In other words, within this same group of transforming miners, Cipher is the most expensively valued yet the least proven.

  My judgment: a ten-year five-fold rise is not impossible, but it requires five high-difficulty things to hold simultaneously — full delivery, cash-flow pass-through, several-fold pipeline re-signing and expansion, no major dilution, and a valuation-framework switch — with a low compounded probability. And today's price of about $23 already treats "successful transformation" as the default and prices in "continued pipeline realization" as an option — the market isn't buying a margin of safety, it's pre-pricing a whole string of realizations that haven't happened yet. Rising five-fold from this level essentially requires a story that is already full of execution risk not only to come entirely true, but to come true in excess.

  Jun 10, 2026
• QWhy hasn't the market recognized all of this yet? Does it not understand, not respect, or not see far enough? What would become the "narrative inflection point"?3/10

  A

  This question needs to be answered in reverse for Cipher: the market hasn't "failed to recognize" its value — it has already fully, even excessively, recognized it. CIFR is the most expensively valued name among comparable transforming miners, and it's up more than 20% over the past month alone. What hasn't actually been fully priced by the market isn't the upside story — it's the downside risks of "delivery/dilution/leverage." So for Cipher, the key "narrative inflection point" isn't a positive catalyst waking the market up — it's some execution crack that wakes the market up from its high-multiple "data-center platform" dream.

  First, let's use the data to prove "the market has long since understood the upside." This Baillie Gifford question presumes a cognitive gap of "not understanding / not respecting / not seeing far enough," but Cipher's price action is exactly the opposite. I verified the current pricing: CIFR at about $23.04, market cap about $9.42 billion (2026-06-09), with market cap up about 42% over the past 30 days. The report calculates P/B of about 10.7x at $18.80; recalculated at today's price, it's already about 13x; against IREN's P/B of about 7.8x and MARA's about 2.2x, Cipher is the most expensive of this group — and it happens to be the least proven (IREN already has 810MW operational and 2.1GW under construction, while Cipher's HPC rent hasn't contributed a dollar to the financials yet). The market clearly hasn't "disrespected" it — quite the opposite, it has given it the highest premium among peers. So the premise of "why hasn't the market recognized it" does not hold for Cipher.

  So what exactly is the market "not seeing"? It's the downside, not the upside. Several things the report keeps emphasizing are precisely what's being downplayed in today's feverish pricing: first, the filing reality — as of Q1 2026 the company still has only one operating segment, Bitcoin Mining, with Q1 revenue of only about $35 million, roughly halved sequentially due to Black Pearl's retirement — a clear mismatch between the valuation narrative and filing reality. Second, the asset-based approach offers almost no cushion — the report estimates book value per share of only about $1.76; if delivery fails, the tail of the share price converging toward book value runs deep. Third, leverage and dilution — total debt principal of about $5.2 billion plus about 144 million shares of potential dilution from two in-the-money convertible tranches and Google warrants (about a 35% increase relative to roughly 405 million shares). Amid the excitement of the AI-infrastructure narrative, the market prefers to fixate on "future anchors" like the 700MW and roughly $11.4 billion in contracted revenue, while selectively downplaying the fact that this value is still stuck in contracts and on construction sites, not yet settled into free cash flow.

  Placing this precisely within Baillie Gifford's "not understanding / not respecting / not seeing far enough" framework: Cipher's cognitive gap runs in the opposite direction. The market doesn't fail to understand it (everyone can tell the AI data center story), doesn't disrespect it (it has awarded the highest P/B), and doesn't fail to see far enough (if anything, it sees too far — discounting ten-years-out NOI back to today already). Its bias is "excessive optimism" — under-discounting execution risk, dilution risk, and delivery lags. This is the mirror image of the "undervalued, not-yet-recognized great growth stock" Baillie Gifford favors: Cipher is more like a transformation story that has been "fully recognized, and even overextended."

  What would become the narrative inflection point? Because the market is already positioned on the optimistic side, an inflection point is more likely to be a downside trigger than an upside catalyst. The report has actually already laid out both directions of inflection points clearly:

  • Upside inflection points (reinforcing the "data-center platform" narrative): Barber Lake and Black Pearl delivered on time, with post-commencement NOI matching management's presented path; converting the signed 700MW into distributable cash flow without material additional dilution; signing further high-quality investment-grade long-term leases in the pipeline. These would make the market more comfortable applying a high "platform" multiple — but because the price already reflects most of this, the marginal surprise from these positives materializing is limited.
  • Downside inflection points (breaking through the "platform" narrative, reverting to "cyclical miner" pricing): campus construction delays, rent commencement pushed back, changes to tenant credit or backstop conditions, another large low-priced equity issuance, deterioration in project-debt terms, or a clear cooling in AI data-center demand. If any of these occur, the market could quickly revert from about 13x P/B platform pricing to something closer to IREN's (about 7.8x) or even MARA's (about 2.2x) miner pricing — and the downside gap in between is enormous.

  My judgment: for Cipher, the honest answer to this Baillie Gifford question is "the market hasn't just recognized it — it may have overpriced it" — it is the most expensive, least-proven transformation name among peers, with a cognitive gap running in the negative direction. What truly hasn't been fully priced is downside risk (delivery lag, dilution, leverage, zero cushion from the asset-based approach). So its key narrative inflection point isn't "the market finally seeing the upside" — it's "the first execution crack appearing and shaking the market awake from its high-platform-multiple dream." Until then, today's price is buying a narrative, not a margin of safety.

  Jun 10, 2026