Bloom Energy Q1 FY26 earnings call BE

The headline event was Oracle's announcement (the night before the call) of a new power paradigm for Project Jupiter, a multi-gigawatt AI factory in New Mexico, whose up-to-2.45 GW power block will be 100% Bloom, replacing previously planned gas turbines and backup diesel generators — positioned as one of the largest islanded microgrid power facilities in the world. KR framed Oracle's pivot as driven by community/air-quality/water/noise concerns and by speed and reliability of grid-independent power. He stressed Jupiter is 'not going to be a one-off^📎' and that well more than half of Bloom's current data-center backlog already comes from other hyperscalers, neoclouds and colocation providers, all similarly using no grid, no diesel, no batteries and no turbines.

Management's central strategic message was a shift from lumpy, one-off^📎 capacity additions completed over a year to continuous additions of 'hundreds of megawatts a quarter,' described as an analog dial rather than a digital step function. Current footprint supports 5 GW/year; beyond that Bloom will build new factories as needed via its 'copy exact' model, claiming it can add capacity in months not years and 'never be a bottleneck to our customers.' KR asserted the pace of revenue growth is now set by how fast customers can build greenfield sites, not by Bloom's supply — an explicit rejection of the legacy power industry's multi-year backlog model.

Q1 delivered record revenue of $751.1M, non-GAAP gross margin of 31.5% (+~280 bps YoY), operating income of $129.7M (vs $13.2M) at 17.3% margin, adjusted EBITDA of $143M (~19% margin), and non-GAAP EPS of $0.44 (vs $0.03). Operating cash flow was a positive $73.6M — the first-ever positive first quarter — aided by a step change in profitability, strong collections and customer prepayments to reserve capacity, leaving $2.52B of cash. On the strength of this, FY26 revenue guidance was raised to $3.4-3.8B (80% growth at midpoint), gross margin to ~34%, non-GAAP operating income to $600-750M and non-GAAP EPS to $1.85-2.25.

KR characterized the AI pipeline as 'diverse and robust,' with Oracle just one of several projects using fully grid-independent, all-Bloom microgrids. He framed AI as a 'rinse and repeat' of the commercial-and-industrial playbook, where 70-80% of business comes from repeat customers. Utility-scale interest is emerging for the first time as favorable rate-base regulation lets both gas and electric utilities partner with Bloom, and reshoring of factories to America was cited as an additional demand vector. The C&I business was described as robust and growing but overshadowed in commentary by the scale of AI deals.

Responding to Evercore, KR agreed most current backlog is tied to AI training but argued inference will be 'much bigger than training in terms of total gigawatt need,' distributed at the edge near dense populations rather than in multi-gigawatt remote data centers. He used a Rhode-Island-scale analogy — a ~2.5 GW CCGT would consume roughly a million showers' worth of water per day and emit car-equivalent pollution — to argue that community resistance to combustion power makes clean on-site generation essential, and even more so for edge inference sited near cities.

KR repeatedly emphasized Bloom's 'genius of and' — clean and reliable and fast and affordable. The servers do not combust, use minimal water (only at startup, none in normal operation), are compact, and meet air-quality requirements in virtually all jurisdictions, making them a better 'neighbor' as permits and community acceptance become the gating factor for AI infrastructure. On cost, over a decade of double-digit reductions has made the servers cost-competitive with grid power in most U.S. markets and with off-grid alternatives in nearly all markets, positioning Bloom as the only on-site solution with a sustained downward-sloping cost curve.

Service revenue was $61.9M (+15.6% YoY) at 18% margin (+13 points YoY), marking a fourth consecutive quarter of double-digit service gross margin and ninth consecutive quarter of service profitability. KR underscored a 100% attach rate between product sales and service — 'not a single deal' without it — with data-center service contracts averaging 10 to 15 years, framed as a 'tremendous source of annuity revenue' expected to keep benefiting from scale and field-performance improvements.

Simon Edwards, formerly CFO at Grok among software franchises, joined as CFO roughly two weeks before the call. KR framed the deliberate year-long search around finding a leader with a systems-engineering background who has scaled manufacturing and applied data/automation-driven operating models. Edwards cited AI-infrastructure tailwinds, real power-availability bottlenecks, Bloom's leadership depth, and the chance to build a 'generational company' as reasons for joining, and emphasized a bias toward converting demand into delivered systems, cash flow and sustainable performance.