Eight Losses, Then a Billion-Dollar Quarter

On July 28, 2026, Bloom Energy reported that it had sold $1.065 billion of product and service in three months. It was the first time in the company's history that a quarter had cleared a billion dollars. Revenue was up 165.5 percent on the same quarter a year before. Product revenue alone was up 215.4 percent.

The figures further down the release were the ones that had never appeared before. Operating income of $182.2 million, against an operating loss of $3.5 million a year earlier. Net profit to common stockholders of $196.3 million, against a $42.6 million loss. Cash from operating activities of $226.4 million, a swing of $439.5 million from the $213.1 million the company had consumed in the same quarter of 2025.

Now the other column.

From 2016 through 2023, Bloom Energy lost money every single year: $279.7 million, $262.6 million, $241.8 million, $304.4 million, $157.6 million, $164.4 million, $301.7 million, $302.1 million. Eight consecutive annual net losses, $2.014 billion in total, incurred while KR Sridhar - founder, chairman and chief executive, all three at once - went on selling the same argument to a market that did not need it yet.

The Product Was Never The Problem

Sridhar came to the company from a laboratory rather than an operating role. He was director of the Space Technologies Laboratory at the University of Arizona and a professor of aerospace and mechanical engineering there, and has advised NASA. His bachelor's degree in mechanical engineering, taken with honours, is from the University of Madras, now NIT Trichy; his master's in nuclear engineering and his doctorate in mechanical engineering are from the University of Illinois at Urbana-Champaign. He has since sat on the board of C3.ai and served as a strategic limited partner at Kleiner Perkins Caufield & Byers.

What Bloom builds is a solid-oxide fuel cell that generates electricity on the customer's own site, next to the load, without combustion. For most of two decades that was a premium answer to a question nobody urgent was asking. American electricity was cheap, the grid was there, and an interconnection took as long as it took.

The engineering worked. The market did not care. That is what $2.014 billion of losses buys you: time.

The Grid Ran Out Of Room

The variable that changed was not the fuel cell. It was the queue.

AI data centres need large blocks of firm power on timelines that utility interconnection cannot meet, and the shortage turned an environmental proposition into a scheduling one. Sridhar's own framing of the shift, in the fourth-quarter results of February 5, 2026, was blunt about the reclassification: "Bring-your-own-power has shifted from a slogan to a business necessity for AI hyperscalers and manufacturing facilities." He added four words that read as a claim on the next decade: "This shift is secular and growing."

The 2025 numbers had already begun to show it. Full-year revenue of $2.024 billion, up 37.3 percent on $1.474 billion in 2024. Operating income of $72.8 million against $22.9 million. A second consecutive year of positive free cash flow. And the disclosure that matters more than any single quarter: total current backlog of approximately $20 billion, with product backlog of roughly $6 billion, up about two and a half times year over year.

Twenty billion dollars of backlog against two billion dollars of annual revenue is not a demand problem. It is a manufacturing problem.

The Quarter That Went Backwards

The record year contained a bad quarter, and the release did not hide it.

In the fourth quarter of 2025, gross margin fell 7.5 percentage points year over year, to 30.8 percent. Fourth-quarter operating income fell $17.2 million year over year, to $87.5 million. The company was booking record annual revenue and record annual gross margin while its most recent three months showed margin compression against the prior year - the signature of a business buying capacity ahead of the volume it has sold.

That is the risk the backlog creates. A company with $20 billion of commitments and $2 billion of revenue has to spend before it collects, and the fourth quarter is what that looks like on the way through.

What He Claims Now

Six months later the margin question was answered, at least for one quarter: 33.4 percent gross margin in the second quarter of 2026, up 668 basis points year over year, on nearly three times the revenue.

Sridhar's account of why is a claim about who has already signed off. Demand keeps accelerating, he said on July 28, as "customers who traditionally defaulted to combustion technologies are now proactively choosing Bloom as a superior power solution." Then the specific assertion, which is checkable and which he chose to put in an SEC-filed document: "all the major US hyperscalers and over a dozen US neoclouds, AI labs, and colocation data center operators have validated and approved our power solutions for their AI factories."

His conclusion was five words long. "Bloom is now a standard for AI onsite power."

The company raised full-year 2026 guidance to $3.9 billion to $4.2 billion of revenue - roughly double 2025 at the midpoint - with non-GAAP operating income of $800 million to $900 million and non-GAAP earnings of $2.55 to $2.85 a share.

The Sentence He Kept

In March 2020, announcing a new chief financial officer and six years away from a profitable quarter, Sridhar described the task ahead as continuing "to execute on our mission to make clean, reliable, and affordable energy for everyone in the world."

He was then running a company that would lose another $157.6 million that year, and $164.4 million the year after, and more than $300 million in each of the two years after that.

The mission sentence did not change. The grid did. Bloom Energy now has to build against $20 billion of orders taken on the strength of an argument its founder spent eight loss-making years making before anyone was ready to buy it.