Bloom Energy In-Depth Value Investment Research

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

• QHow large is its market ceiling? Is it expanding an existing pie, or creating an entirely new market?6/10

  A

  Bottom line: the ceiling is high enough, but Bloom is mainly expanding an existing pie that AI is repricing, namely power supply for data centers, rather than creating a market from scratch. The truly incremental demand is the niche for rapid onsite power forced into existence by grid bottlenecks. The runway is real, but Bloom still has to fight gas turbine giants for its share.

  Start with the size of the "pie" itself. Bloom defines itself as a global leader in onsite power generation, serving data centers, semiconductors, AI infrastructure, utilities, and industrial loads. It is not selling a new need. Supplying power to loads has existed since the birth of the electric power industry. What it has captured is a structural change on the demand side: AI compute expansion has run into delayed grid expansion, so time-to-power has moved from a secondary metric to the first constraint in vendor selection. The report cites company disclosure that, within certain project scopes, Bloom can deliver onsite power in about 90 days. That is exactly the opening it is exploiting.

  The reality of demand is backed by external orders, not PPT slides. The three partnerships listed in the report have all translated into specific dollar amounts: the AEP 1GW supply framework corresponded to about a $2.65 billion contract under the disclosure basis for Q1 2026; Brookfield announced a five-year project financing framework of up to $5 billion in October 2025; and the Oracle partnership also included warrants for up to 3,531,073 shares with an exercise price of $113.28. All of this points to the same conclusion: the onsite data-center power pie is growing rapidly, and major customers are willing to commit real money and long-term frameworks to it.

  But a high ceiling does not mean Bloom owns it all. From Baillie Gifford LTGG's question of "how high is the market ceiling," the honest answer requires separating two layers. The TAM, data centers plus industrial onsite power, is indeed a multi-hundred-GW long runway. But Bloom's SAM, the share it can actually capture, is constrained because it must compete with gas turbine and large power-equipment players such as GE Vernova and Siemens Energy for the same data-center power budgets. Barron's explicitly noted that Bloom's target today is no longer just the fuel-cell niche, but a fight for share in the data-center power market against traditional large equipment suppliers. This is a story of taking share in a huge and still expanding existing market, not of defining a previously nonexistent category that only Bloom can supply.

  So on the binary question of expanding an existing pie versus creating a new market, the honest answer leans toward the former. Bloom has repackaged solid oxide fuel cells from a "clean energy supplement" into a "fast solution to the AI power bottleneck." That is a resegmentation and upgrade of the existing power-supply market, not market creation out of thin air. The truly new part is packaging equipment, installation, financing, tax incentives, and long-term operations and maintenance into an onsite-power-as-a-service delivery model. That lowers the adoption threshold and expands addressable demand, but the underlying business remains the old business of supplying power to loads. The height of the ceiling is not in doubt. How large and durable a piece Bloom can hold under that ceiling is the real uncertainty the next 9 questions need to answer.

  Jun 10, 2026
• QCan its revenue at least double over the next five years? Will growth mainly be driven by volume, price, or new businesses?7/10

  A

  Bottom line: doubling revenue over the next five years is almost a low bar. By management's own guidance, 2026 revenue could jump from about $2.02 billion in 2025 to $3.4 billion–$3.8 billion, nearly doubling in a single year. Growth is mainly driven by volume, namely higher product shipments, with price and new businesses secondary. But doubling is not Bloom's real challenge. The hard part is whether, after doubling, it can turn volume into sustainable cash flow without heavy dilution.

  First set out the base and growth rate. Bloom reported about $2.02 billion of full-year 2025 revenue, up 37.3% year over year; product revenue rose from $1.09 billion in 2024 to $1.53 billion in 2025, up 41.1%. Growth steepened further entering 2026: Q1 2026 revenue was $751 million, up 130.4% year over year, including product revenue of $653 million, up 208.4%. Management accordingly raised full-year 2026 revenue guidance to $3.4 billion–$3.8 billion. In other words, "doubling in five years" is almost a false question for Bloom. It is close to doing that in one year; achieving it in five years is the base case, not the challenge case.

  What is driving growth? Mostly volume. Product revenue growth of 208.4% year over year corresponds to shipment growth in fuel-cell systems, not a jump in unit prices. The report also notes that the company still has to offer commercial concessions to win key customers, including Oracle warrants and embedded derivative clauses under the power-price-based Escalation Protection Plan in some contracts. That shows Bloom does not have absolute pricing power, leaving limited room to drive growth through price increases. Capacity supports the volume ramp: the company plans to double manufacturing capacity from 1GW to 2GW by the end of 2026, and the report adds that each additional 1GW of annual capacity requires roughly $100 million–$150 million and about 6–9 months. The growth formula is therefore data-center orders, capacity expansion, and larger shipments. Volume is the clear lead driver.

  Price and new businesses are supporting actors. On price, as noted, there is no structural pricing power. On new businesses, Bloom Electrolyzer is theoretically a second product category, but the report clearly discloses that the company has impaired first-generation electrolyzer assets and stopped promoting that generation. Electrolyzers do not constitute a near-term growth engine. Service revenue is growing, about $228 million in 2025, but it remains small, and the report notes that operations and maintenance contracts, though as long as 5–20 years, can be terminated for convenience each year, so stickiness is lower than it appears. Revenue doubling over five years is essentially a volume ramp on the main product-equipment curve, not a multi-engine advance.

  But to answer this Baillie Gifford question honestly, one must point out that "doubling" is not Bloom's real test. The core tension emphasized repeatedly in the report is the quality of growth. In 2025, the top three customers contributed about 43%, 13%, and 12% of revenue, about 68% in total, and much revenue came through Fund JV structures related to Brookfield. This means the doubling is driven by a small number of major customers, financing structures, and large-project volume. It can be powerful on the way up, but fragile if customers or financing partners change course. The cost of doubling is also dilution: the report records about 240 million diluted shares in FY2025 and about 264 million TTM, and the company also issued $2.5 billion of 0% convertible notes in 2025. So the honest answer is: revenue doubling within five years is almost certain and volume-driven, but that is not the scarce capability worth extra credit. What remains to be proven is whether the doubling can occur without reliance on continuous financing and share-count expansion. That will determine how much shareholders ultimately receive.

  Jun 10, 2026
• QFive years from now, what will take over as the next growth engine? Does this "second curve" exist today?3/10

  A

  Bottom line: Bloom does not currently have a formed, visible "second curve." All of its growth is still staked on the main curve of onsite power-equipment volume. Theoretical succession candidates, hydrogen electrolyzers, sticky long-term service/power revenue, and international markets, have either been halted by the company, remain too small, or are extensions of the same curve. For Baillie Gifford's question of what takes over five years from now, the honest judgment is that the second curve does not yet exist. This is a real gap in Bloom's long-term story.

  First exclude the easiest candidate to mistake for a second curve: electrolyzers. Bloom nominally has two product lines, Energy Server for power generation and Electrolyzer for hydrogen production. But the report clearly discloses that the company has impaired assets related to the first-generation electrolyzer and stopped promoting that generation of products. A product line already impaired and discontinued by the company itself plainly cannot count as a second growth engine that exists today. Hydrogen may restart in the future, but at present it is a shelved option, not a running curve.

  Next look at service and power revenue as the candidate for "utility-like cash flow." This has the greatest second-curve potential, but it is far from taking over today. Among the company's four revenue categories in 2025, product revenue of $1.53 billion was the dominant piece, while service was about $228 million and power was about $60 million. Service plus power was less than 15% of total revenue. More important is quality: the report notes that maintenance and operations contracts, though as long as 5–20 years, can be terminated for convenience each year; and in 2025, 98% of revenue came from direct purchase arrangements, with only 2% from managed services. This shows that Bloom remains driven by equipment transactions. The theoretically more attractive second curve, an installed-base flywheel compounding through spare parts and O&M annuities, is still barely formed.

  Do international expansion and new use cases count as a second curve? They count as extensions, not a new curve. The company had deployed about 1,100 sites across 9 countries by the end of 2025. Geographic expansion and expansion from data centers into semiconductors and industrial customers are mostly replication of the same product across more customers. The driver remains equipment volume. In substance this is still the main curve, not an independent new growth pole that does not depend on the existing product's success.

  So the implicit premise of this Baillie Gifford question, "does the second curve exist today," must be answered directly: no independent, formed second curve exists. Bloom's current position is that the main curve, data-center power equipment, is rising steeply, while the second curve has not yet ignited. This is a valuation demerit. The report's intrinsic value range, $45–$75 in the base case and $90–$130 in the optimistic case, already offered no margin of safety against the roughly $259 current share price, while the story supporting the current price implies not only a perfect main curve but also high growth sustained for many years. When a company lacks a second curve and is priced as a long-term big winner, any slowdown in the main curve has no buffer.

  A balancing point is worth adding: the absence of a second curve does not doom the company to mediocrity. If Bloom later turns the installed base into sticky service/power annuities, meaning that 2% managed services share rises meaningfully, or if hydrogen restarts after policy and cost inflection points, a second curve could grow later. The report also lists "a further shift in the revenue mix from equipment transactions toward sticky long-term earnings" as one of the key facts that would significantly raise its view. But "could grow later" and "already exists today" are different claims. By Baillie Gifford's standard of respecting reality and not elevating narrative, this box should be marked as empty today.

  Jun 10, 2026
• QWhat is its core competitive advantage? Will this moat widen or narrow over the next three to five years?5/10

  A

  Bottom line: Bloom's core competitive advantage is not brand or absolute cost. It is the combination of deployment speed, modularity, onsite availability, and product fit for data-center use cases, plus a financing and channel capability built through relationships with SK ecoplant, AEP, Brookfield, Oracle, and others. But this moat is currently "starting to show, still under construction." Whether it widens or narrows over the next three to five years is highly uncertain. It depends on whether Bloom can turn onsite power into an industry standard before gas turbine giants close the "rapid power" gap.

  Start with the real source of advantage. Bloom's differentiation is time-to-power. The report discloses that, within certain project scopes, Bloom can deliver onsite power in about 90 days. In a setting where AI data centers are scrambling for power and grid expansion can take years, that has hard value. Together with modularity, scaling while building, high availability, described by the company as "ultra-reliable," and low-emission power without combustion, it serves customers that are highly sensitive to power timing and onsite deployment. This capability has been validated by major external orders: the AEP 1GW framework, about $2.65 billion, Brookfield's five-year financing framework of up to $5 billion, and the Oracle partnership and warrant arrangement. These all show that large customers recognize this delivery capability.

  Then break down the moat sources honestly, item by item. Brand: moderately weak, with enterprise recognition but no consumer brand and no clear strong pricing power. Cost advantage: unclear. Gross margin has improved to about 29% in 2025 and GAAP 30.0% in Q1 2026, but that is not enough to prove structurally low cost. Scale: moderate. Capacity is planned to rise from 1GW to 2GW by the end of 2026, and further expansion could spread unit costs. Switching costs: moderate. Deployed systems, operations and maintenance, and site adaptation create some stickiness, but the report also notes that O&M contracts can be terminated for convenience each year. Network effects: weak; this is not a platform business. Patents: 380 issued U.S. patents and 252 international patents as of year-end 2025, but the company explicitly says the business does not depend on any single patent, so patents are more defensive than monopolistic. This is a moat made of several medium-strength sources layered together, with no single source strong enough to be wide and deep on its own.

  The key counterweight: it does not have absolute pricing power. The report notes that the Oracle deal included warrants for up to 3.531 million shares at an exercise price of $113.28, and that some sales contracts include power-price-based embedded derivative clauses under the Escalation Protection Plan. Bloom still has to offer commercial concessions to win key customers. A company with a truly wide moat says, "this is the price, and you have to pay it." Bloom says, "I can win orders through speed and fit, but I must give up part of the economics." That separates it from great companies with strong pricing power.

  So will the moat widen or narrow over three to five years? This is a genuinely open race, and I agree with the report's judgment. If Bloom turns onsite power for AI data centers into a de facto standard, lifts service and spare-parts margins sustainably, and converts capacity into scale cost advantages, the moat will widen. Conversely, if large gas turbine and power-equipment players such as GE Vernova and Siemens Energy catch up on deployment speed, interconnection coordination, and total cost of ownership, the "speed" Bloom now relies on will be offset, and the advantage will narrow quickly. In other words, its moat protects a time window. The width of that window is how long competitors remain unable to close the rapid-power gap.

  Overall judgment: this is a strong participant that found a high-growth opening in an ordinary industry. The moat is real but not yet formed, and its direction is uncertain. By Baillie Gifford's reality-based standard, today's moat should be scored as medium and uncertain in future width. It should not be elevated to "already wide" because of the AI story, nor should Bloom be dismissed merely because it is still GAAP net-loss-making, since it has already built advantages in speed and channels.

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

  A

  Bottom line: Bloom shows some signs of self-reinvention. It has repositioned a long-loss-making clean-energy concept company as a rapid-power supplier to AI data centers, and it decisively impaired and stopped promoting its first-generation electrolyzer after that product failed. That shows a willingness to acknowledge mistakes and adjust direction. But this reinvention is more a change of narrative and use case on the same technology base than hard evidence of being able to rebuild from scratch after its core is disrupted. On mistakes and bad news, disclosures are relatively candid, while the aggressive capital structure also reveals an offense-first culture.

  First add the implicit premise of this question: the "DNA for self-reinvention" should be judged on two things. Can the company switch arenas and survive if the core business is disrupted, and how does it normally handle mistakes and bad news, which determines whether you can trust its self-reporting. The two should be assessed separately.

  Start with evidence of rebirth through a new use case. The same solid oxide fuel-cell technology that Bloom previously framed as a clean-energy/decarbonization story has, over the past two years, been successfully repositioned as a solution to the AI power bottleneck, with product value anchored to time-to-power, about 90 days to power. This repositioning has translated into results: Q1 2026 revenue was $751 million, up 130.4% year over year, and product revenue grew 208.4%, alongside major orders such as AEP 1GW, about $2.65 billion, Brookfield's five-year financing framework of up to $5 billion, and the Oracle partnership. Turning the business from "decarbonization equipment waiting for policy support" into a "power platform chased by compute demand," without changing the core technology, is itself a form of narrative and business-model reinvention.

  Next look at how it handles its own mistakes, which is the key integrity test for the reinvention DNA. The hardest evidence is the electrolyzer: the company has impaired first-generation electrolyzer assets and stopped promoting that generation of products, as disclosed in the report. This is an impairment that admits, "the first generation of this route did not work," rather than forcing it through or dressing it up. Willingness to write down a failed product line and refocus on the most profitable power-supply core is healthy corrective behavior. This fits the report's judgment: the company gives fairly full 10-K disclosure on financing dependence, installation delays, policy risk, complex JV structures, and reliance on accounting estimates, and Deloitte issued an unqualified opinion on the 2025 financial statements and internal control. There is no obvious sign of avoiding bad news or glossing over everything. Its handling of bad news is relatively candid, which makes the reinvention story more credible.

  But two limits to the "reinvention DNA" need to be stated honestly. First, this reinvention is at the use-case/narrative level, not the level of building a new house after the core is disrupted. Bloom has never faced a life-or-death test in which its core technology, solid oxide fuel cells, was fully displaced by a superior solution and the whole company had to transform. It has proved that it can change the story and customer scenario, but it has not proved that it can rebuild a new core if the existing one is overturned. Second, the cultural baseline is offensive rather than conservative. The report records that the company issued $2.5 billion of 0% convertible notes in 2025 and exchanged old debt, recognizing about $32.3 million of early extinguishment losses; dilution is heavy, buybacks are almost nonexistent, and the short-term incentive bonus pool can reach 200% after targets are met. This culture of willingness to finance, dilute, and pay high incentive upside is ammunition for reinvention in favorable conditions, but it can also magnify risk in adverse ones.

  Overall judgment: Bloom has early reinvention DNA in the form of willingness to adjust, cut losses, and disclose candidly. Its handling of mistakes and bad news is relatively healthy, which is a plus. But it has not yet passed a true core-disruption exam. Its reinvention capability remains mostly at the change-the-use-case level, and its offensive capital culture is a double-edged sword. By Baillie Gifford's standard of not overstating the case, this dimension is positive but untested. It deserves credit for corrective integrity, but one successful narrative shift should not be misread as hard DNA for surviving disruption.

  Jun 10, 2026
• QDoes management, especially the founder, have a long-term view and deep alignment with the company? Is it willing to sacrifice current profit for five to ten years from now?5/10

  A

  Bottom line: founder KR Sridhar is the technical and strategic core of the company, with a clear long-term view and clear willingness to spend for the long term, shown by years of losses while continuing in-house R&D and by aggressive financing to exchange for growth. But his alignment with the company is medium rather than deep. A stake of about 2.0% is not a heavy personal bet, and the company's aggressive dilution and offensive compensation structure are not the most shareholder-friendly capital allocation culture for long-term owners.

  Start with long-term view and the willingness to sacrifice the present for five to ten years out. Sridhar passes on this point. He is the founder, chairman, and CEO, and is plainly the central figure in the company's history and product evolution. Bloom spent more than a decade continuously investing in solid oxide fuel cells while enduring long periods of losses. The report discloses large consecutive net losses from 2021–2023, with an operating inflection only appearing in 2024. That itself is behavioral evidence of willingness to sacrifice current profit for long-term payoff. In the past two years, the company has also dared to use large-scale financing to convert capacity and orders into revenue acceleration, turning the technology story into a revenue surge. The execution of long-termism is improving. This is a trait Baillie Gifford would appreciate: willingness to tolerate long non-profitability to build something.

  But "deep alignment with the company" needs an honest discount. The 2026 proxy materials disclose that, as of February 27, 2026, Sridhar held about 5.662 million shares, about 2.0% of outstanding Class A common stock; management and directors together held about 3.0%. This is not a zero-share professional manager, but it is far from a founder whose net worth is fully tied to the company. For a founder-CEO, 2.0% is not high. The company does have executive stock ownership guidelines, requiring the CEO to hold 4 times annual salary, which is directionally right, but 4 times salary is only a symbolic constraint relative to a market capitalization of about $7.38 billion. Compared with founders who own 20%–30% and whose interests are tightly aligned with minority shareholders, Sridhar's alignment is "present, but not deep."

  The more important test is whether capital allocation culture is friendly to shareholders. I agree with the report's judgment: dilution is heavy, debt is large, and buybacks are almost absent. In 2025, Bloom issued $2.5 billion of 0% convertible notes, and in the November debt exchange issued 24.30 million and 18.11 million Class A shares to holders of 2028/2029 convertible notes. Diluted shares were about 240 million in FY2025 and about 264 million TTM. The report records share-count changes of about +5.73% in FY2025 and about +7.50% TTM. In compensation, 2025 short-term incentives were centered on total revenue and non-GAAP operating profit, the bonus pool could reach 200% after targets were met, the CEO's 2025 cash incentive was about $2.47 million, and the compensation committee made additional adjustments to non-GAAP operating profit for bonus purposes. This combination of willingness to dilute, high bonus elasticity, and incentives anchored in growth rather than per-share intrinsic value is a classic offensive growth-stock culture. It is not outrageous for a growth company, but from a long-term owner's perspective, the constraints on per-share value, capital returns, and dilution control are weak.

  The integrity baseline is a plus and should be credited. The report notes that the company gives fairly full 10-K disclosure on financing dependence, policy risk, complex JV structures, and reliance on accounting estimates, and Deloitte issued an unqualified opinion. There is no obvious sign of dressing up or avoiding bad news. This is not a fraud team; it is a technically driven team with stronger financing capability, but a clearly aggressive growth-stock style in the use of capital.

  Overall judgment: the long-term view and willingness to spend for the long term are strong and pass the test; honesty is neutral to positive; but alignment is medium, with 2.0% ownership, and capital allocation is not especially friendly to long-term shareholders, given heavy dilution and incentives anchored in growth. By Baillie Gifford's standard, this dimension is "trustworthy but not ideal." Sridhar looks like a founder you can follow, but whose dilution habits must be watched closely, not an owner-CEO whose capital discipline and deep alignment let you fully relax.

  Jun 10, 2026
• QIf it disappeared tomorrow, how much would customers miss it? Is its growth sustainable and not dependent on harming society or regulation?5/10

  A

  Bottom line: if Bloom disappeared tomorrow, the data-center customers most dependent on it would miss it materially. The roughly 90-day onsite power it provides is a scarce solution in today's grid bottleneck and is not easy to replace seamlessly in the short term. But its indispensability is situational and can be approached by substitutes, rather than absolute and exclusive. The sustainability of growth has real concerns: heavy dependence on tax incentives, financing structures, and a few major customers. It does not harm society, but its dependence on regulatory and policy framing is high.

  First add the double premise of this question: we need to assess both indispensability, meaning how painful and hard to replace it would be if it disappeared, and whether growth is sustainable without depending on harm to society or regulation. The two should be judged separately.

  Start with indispensability. The conclusion is strong within the use case, medium in absolute terms. Bloom's core value proposition is time-to-power. In the window where AI compute expansion has hit delayed grid expansion, it can deliver onsite power in about 90 days, as disclosed in the report. For data-center customers scrambling for power and unable to wait for large grid expansion or gas turbine delivery schedules, that is a hard need. Its stickiness also comes from deployed-system O&M, site adaptation, and switching costs created by customer validation, as well as major customers' real-money commitments: the AEP 1GW framework of about $2.65 billion, Brookfield's five-year financing framework of up to $5 billion, and the Oracle partnership. If these customers lost Bloom, it would indeed be hard in the short term to find an equally fast, onsite, modularly scalable substitute.

  But indispensability has a ceiling, and this must be stated honestly. The report notes that Bloom must compete with gas turbine and large power-equipment players such as GE Vernova and Siemens Energy for the same data-center power budgets. That means it is not the only company that can power data centers; it currently leads on rapid deployment. Once gas turbine or grid-expansion solutions catch up on speed and total cost of ownership, customer "miss" intensity would cool quickly. It also lacks absolute pricing power, as shown by Oracle warrant concessions and power-price-related embedded derivative clauses in contracts, meaning customers have bargaining power and are not helpless without Bloom. So the honest answer to how much customers would miss it tomorrow is: locked-in existing projects would hurt badly, new demand would find alternatives, and the pain would be strong but not existential.

  Now consider growth sustainability and the social/regulatory side. On not harming society, Bloom is positive: its product is onsite power without combustion and with low emissions, directionally helpful for emission reduction. It is not making money by harming users or imposing social externalities. The sustainability concern is dependence on policy and structure. The report is clear that the company's business is heavily affected by power pricing, interconnection rules, tax incentives such as ITC and accelerated depreciation, environmental standards, and subsidies at multiple levels of government. In its 10-K risk disclosure, the company itself acknowledges that changes in tax incentives and the financing environment could harm demand, financing, and profitability. Add that the top three customers account for about 68% of revenue, and that much revenue comes through Fund JV structures related to Brookfield. This growth driven by policy benefits, financing structures, and a few large customers can be very powerful on the way up, but its sustainability depends on external conditions remaining favorable. This is not harming regulation; it is excessive dependence on regulatory and policy friendliness, a source of fragility.

  Overall judgment: indispensability is high within the scenario, supported by the scarcity of roughly 90-day power and installed-base stickiness, but it is not absolutely exclusive and can be approached by gas turbine/grid solutions. The growth model does not harm society, given low emissions and positive externalities, but its dependence on tax incentives, financing structures, and major customers is high, so sustainability depends on a friendly policy and capital environment. By Baillie Gifford's standard, this dimension is "moderately positive but conditional." Customers would miss it, but that attachment rests on a leading window that has not yet been tested by competition and policy cycles.

  Jun 10, 2026
• QWhat are the unit economics of this business, including gross margin and incremental returns? Do they improve or deteriorate with scale? Where does the money it earns go?5/10

  A

  Bottom line: Bloom's unit economics are moving from ugly to acceptable. Gross margin has recovered from 12.4% in 2022 to about 29% in 2025 and GAAP 30% in Q1 2026, while operating margin and ROIC have both turned positive. Directionally, scale is making the economics better. But incremental returns have not yet been proven across a cycle, and much of the money earned is consumed by capacity expansion and working capital rather than settling into distributable free cash flow for shareholders. SBC also continues to dilute per-share value.

  Start with the trend in gross margin and operating leverage. This is real improvement. The report and primary data are consistent: gross margin was 20.3% in 2021, 12.4% in 2022, the trough, 14.8% in 2023, 27.5% in 2024, and about 29% in 2025, while Q1 2026 GAAP gross margin was 30.0% and non-GAAP was 31.5%. Operating margin improved from -21.8% in 2022 to about 3.6% in 2025 and about 6.7% TTM; Q1 2026 GAAP operating profit was $72.19 million, and non-GAAP operating profit was $129.7 million. This shows that as shipments scale, fixed costs are being spread and operating leverage is starting to appear. The direction of unit economics is clearly improving. This is the core evidence that Bloom is graduating from "always almost there." Management's 2026 full-year non-GAAP gross margin guidance of about 34% further points to scale improving unit economics.

  Incremental returns, ROIC, have also turned positive, but they are not yet thick enough. Under the report's basis, ROIC improved from -23.2% in 2022 to about 5.0% in 2025 and about 8.1% TTM. That is real progress, but ROIC just above 8% is only the beginning of value creation for a company priced as a long-term big winner. It is still far from excellent capital returns. The report's investment checklist directly marks "whether capital returns are excellent" as a fail. Whether scale can push ROIC to structurally high levels remains unproven.

  The key honest point is where the money earned goes: much of it is consumed by capacity and working capital, and does not settle into shareholder free cash flow. The report shows FY2025 operating cash flow of about $114 million and free cash flow of about $57 million; by TTM through 2026Q1, operating cash flow was about $298 million and FCF about $230 million, so cash flow is indeed improving. But quality needs unpacking: positive 2026Q1 operating cash flow came in meaningful part from about $93.10 million of increases in customer advances and deferred revenue, while inventory increased about $88.60 million and receivables plus contract assets increased about $51.90 million. That shows Bloom is in a stage where high growth consumes working capital, and cash flow is not mature enough to ignore project timing. Capacity expansion is the other major destination for money: capacity is planned to rise from 1GW to 2GW by the end of 2026, with each additional 1GW requiring about $100 million–$150 million of capital expenditure. So the money earned is mainly absorbed by reinvestment and working capital, not returned to shareholders through buybacks or dividends.

  Stock-based compensation, SBC, also needs to be counted as a real cost in unit economics. The report emphasizes that SBC must not be treated as "free": SBC was about $139 million in 2025, and the company did not offset it with buybacks. The November 2025 debt exchange also issued 24.30 million plus 18.11 million Class A shares. Share count changed about +5.73% in FY2025 and about +7.50% TTM. After treating dilution as a real transfer of shareholder value, the report's conservative owner earnings distributable to common shareholders are only in the range of about $30 million–$70 million, far below what accounting FCF appears to show. In other words, unit economics are improving, but per-share unit economics are being dragged down by dilution.

  Overall judgment: gross margin and operating leverage truly improve with scale, which is positive and credible. ROIC has turned positive but remains thin. Money is mainly spent on capacity expansion and working capital, has not yet settled into distributable shareholder cash, and SBC continues to dilute per-share value. By Baillie Gifford's standard, this dimension is "clearly improving, but incremental returns and cash conversion have not yet been proven through a cycle." Bloom is now profitable growth rather than cash-burning growth, but it is not yet a mature, high-quality unit-economics business where scale reliably thickens free cash flow per share.

  Jun 10, 2026
• QWhat conditions must all hold for it to rise fivefold in ten years? Are those conditions realistic? What expectations are embedded in today's share price?2/10

  A

  Bottom line: starting from a current price of about $259 and a market capitalization of about $73.8 billion, a fivefold rise over the next ten years would imply a market capitalization of about $370 billion. That requires four conditions to hold at the same time: sustained high-speed revenue volume growth, a structural rise in gross margin to high levels, valuation multiples staying high, and dilution being meaningfully restrained. The difficulty is extremely high. More importantly, today's share price has already prepaid for a nearly perfect future. Under the report's three-scenario discounting, even the optimistic intrinsic value is only $90–$130, so the current price embeds expectations more aggressive than the report's most optimistic assumptions.

  First translate "fivefold in ten years" into hard metrics. BE closed at about $259.61 on June 9, with a market capitalization of about $73.8 billion. Fivefold means a market capitalization of about $370 billion. To support that market value, at least the following conditions must all be true:

  First, revenue must compound at a high rate for ten years. Starting from about $2.02 billion in 2025 or 2026 guidance of $3.4 billion–$3.8 billion, supporting a $370 billion market capitalization would require revenue to reach the tens-of-billions-of-dollars scale, even under a generous mature multiple. That implies AI/data-center onsite power demand does not fade over ten years and Bloom continues to capture a large share.

  Second, gross margin and operating margin must structurally rise and stay there. Current GAAP gross margin is about 30%, 2026 non-GAAP gross margin guidance is about 34%, and ROIC is about 8%. To support a fivefold rise, profitability has to climb further to a level that can generate thick free cash flow and remain stable across cycles, rather than fluctuate materially with project timing.

  Third, valuation multiples must not contract sharply. This is the harshest condition. The report gives current valuation at about 30 times P/S, 30.5 times EV/Sales, 324 times P/FCF, and 416 times EV/EBIT, with forward P/E of about 112 times. From such a high starting point, another fivefold rise requires either that multiples remain high, which is very difficult because growth-stock multiples usually compress as companies mature, or that earnings growth "grows into" the multiple while the share price still rises fivefold. That requires earnings growth far beyond fivefold to offset multiple reversion.

  Fourth, dilution must be meaningfully restrained. The report records about 240 million diluted shares in FY2025 and about 264 million TTM. The company issued $2.5 billion of 0% convertible notes in 2025, and the TTM share-count change was about +7.50%. If a fivefold market-cap increase is achieved through continuous issuance and dilution, the per-share fivefold return will be heavily impaired. So a fivefold share price also embeds the expectation that the company will no longer rely on severe dilution for financing.

  Are these conditions realistic? Taken one by one, none is impossible. But asking them to hold simultaneously, demand not fading for ten years, margins rising and holding, high multiples not collapsing, and dilution being contained, multiplies several independent optimistic assumptions. The joint probability is low. Baillie Gifford LTGG does seek great growth stocks that can rise fivefold in ten years, but the premise is undervaluation the market has not yet recognized. Bloom's problem is the opposite: the market has already priced in its fivefold script.

  This is exactly the expectation embedded in today's share price that must be called out. The report's three-scenario owner-earnings DCF gives intrinsic value of $20–$35 in the conservative case, $45–$75 in the base case, and $90–$130 in the optimistic case. The optimistic case already assumes 2026E owner earnings of $650 million, ten-year compound growth of 22%, and a 9% discount rate. The current price is about $259, roughly twice the $130 upper end of the optimistic case. That means the market is not paying for reasonable growth. It is paying for an earnings path even more aggressive than the report's most optimistic assumptions, plus high valuation multiples that do not compress for a long time. The report therefore estimates that buying at about $259 would produce expected annualized returns of -18% to -10% per year in the conservative case, -8% to 0% per year in the neutral case, and only 3% to 8% per year in the optimistic case. Even if operations continue to improve, ten-year returns from this starting price are likely mediocre or negative.

  Overall judgment: the four conditions required for a fivefold ten-year return are not absurd in isolation, but their simultaneous fulfillment is a low-probability event. The current price already embeds even more aggressive expectations and has prepaid for years of perfect execution. By Baillie Gifford's standard, this dimension is "the upside conditions are unrealistically demanding and already overdrawn by the share price." This is not a cheap ticket to a winner the market has not recognized; it is an expensive entry ticket that the market has already priced as a major winner.

  Jun 10, 2026
• QWhy has the market not recognized all this yet? Is it because investors do not understand it, look down on it, or cannot look far enough? What would become the "narrative inflection point"?3/10

  A

  Bottom line: for Bloom, the Baillie Gifford motif of "the market has not recognized it yet" largely does not apply. The opposite is true: the market has recognized it and may have thought too far ahead. It already prices Bloom as a scarce, high-certainty AI power platform. The real perception gap is not that misunderstanding, neglect, or short-termism makes the stock cheap, but that the market discounts risks around growth quality, dilution, and policy dependence too little. The narrative inflection point is more likely to be downward: any loss of momentum in growth, customers, or policy could puncture the premium.

  Answer the question directly first: why has the market not recognized it? The honest answer is that it already has, and perhaps too much. In one year, BE's share price moved from a 52-week low of $20.93 to a high of $322.83 and is now about $259.61, with a market capitalization of about $73.8 billion and forward P/E of about 112 times and P/S of about 30 times. That is plainly not a stock ignored because investors do not understand it or look down on it. It is a stock heavily pursued under the AI narrative and priced to the brim. The report is precise: the market is not pricing Bloom as an industrial company, but as an extremely strong, extremely scarce, extremely high-growth and high-certainty AI power platform. So for Bloom, this Baillie Gifford question should be inverted: has the market thought too much into it?

  Why say "thought too much" rather than "not yet seen"? Three risks are under-discounted by the market:

  First, growth quality is covered by the optimistic story. The market sees the explosion of Q1 2026 revenue growth of 130.4% and product revenue growth of 208.4%, but growth is highly concentrated. In 2025, the top three customers accounted for about 43%/13%/12%, or about 68% in total, of revenue, and much revenue came through Brookfield-related Fund JV structures. This is growth driven by a few major customers, financing structures, and large-project volume. It is powerful on the way up and fragile on the way down. The market's premium does not fully account for this fragility.

  Second, the cost of dilution and complex capital structure is being ignored. The company issued $2.5 billion of 0% convertible notes in 2025, issued tens of millions of new shares in a debt exchange, still reported a full-year GAAP net loss of about $88.40 million, affected by about $32.30 million of debt extinguishment losses and other items, and had a FY2025/TTM share-count change of about +7.50%. The market cheers operating improvement, but the fact that part of operating success is transferred away through dilution and capital transactions is not reflected in the high valuation it assigns.

  Third, policy and financing dependence are being treated as background noise. The company itself acknowledges in its 10-K that the business is highly dependent on ITC, accelerated depreciation, local subsidies, and interconnection rules, and that changes in policy stance could harm demand, financing, and profitability. During a euphoric narrative phase, this kind of systemic policy dependence is easy for the market to temporarily overlook.

  So the real perception-gap conclusion is that the market is not underpricing Bloom because it cannot look far enough. It is under-discounting risk because it is too excited about what it sees. By Baillie Gifford's standard of looking for cheap great companies the market has not yet recognized, Bloom does not qualify. Its good points are already fully or even excessively recognized, and the cheap margin of safety is absent. The report judges that even versus the relatively optimistic intrinsic value case of $90–$130, the current price around $259 is still a clear premium.

  Now add the implicit question of what could become the narrative inflection point, and state honestly that for Bloom the inflection is more likely downward:

  Downward inflections, more likely and more damaging, include several consecutive quarters revealing dependence on a single or a few projects; free cash flow turning negative again while inventory continues to build; large-customer or financing-partner progress at AEP/Brookfield/Oracle falling short of expectations; large new dilution or high-cost capital transactions; service and product gross margins stopping their improvement; and adverse changes in ITC, interconnection, or other policy. If any one of these is confirmed, the market could reprice Bloom from an AI power platform back to an ordinary growth industrial. The report estimates that if the multiple contracts from above 30 times EV/Sales to ordinary levels, the share price could fall 50%–80%, even if the company remains operationally healthy.

  Upward inflections would not be about being "discovered," but about proving that Bloom deserves the premium it already has. Examples include two or three consecutive years of high-quality, high-conversion free cash flow without abnormal dilution; materially higher service/O&M margins and a revenue shift from equipment transactions toward sticky long-term annuities; and lower customer concentration as demand expands from a few major orders to a broader customer base. These would cause the report to raise its view materially, but their role would be to validate optimism that is already priced in, not to reveal a neglected bargain.

  Overall judgment: for Bloom, "why has the market not recognized it" is the wrong question. The market has already fully priced it, and risk discounting is insufficient. By Baillie Gifford's non-inflated standard, this dimension is "no positive perception gap, full premium, and a narrative inflection skewed downward." It is not a great growth stock waiting to be discovered under dust; it is a stock already under the spotlight, relying on future execution to support a high price. Any loosening of the narrative could become a downward inflection point.

  Jun 10, 2026