Bloom Energy just printed a Q2 that, at first glance, looks exactly how bulls want it. Revenue up nearly 20%. Margins moving the right way. Cash flow, finally stable. The stock has had a monster run over the past year, and the story has fresh fuel from clean hydrogen policy clarity, DOE hub funding in motion, and a surge in data center power demand that is pulling every form of firm, on site generation into the conversation.

When I dig into the details, then layer in the tape, I get a different feeling. The business is healthier, yes. The positioning is bullish, yes. But the price action is being held up by flow, and by an expectation of very clean execution. That is a fragile balance. I will walk you through the financials in plain English, then show you what the options and dark pool data are saying, and why I still think a short term correction is likely unless the macro breaks perfectly bullish.

Revenue landed at $401.2 million, up 19.5% year over year. Management kept full year 2025 guidance at $1.65 to $1.85 billion, which puts a real execution bar on Q3 and Q4. I like that, because it forces discipline into the back half.

Margins stepped up. Non GAAP gross margin came in at 28.2%, GAAP at 26.7%. The spread is narrow enough to trust. On the bridge, GAAP gross profit was $107.1 million, add $6.2 million of adjustments, and you get $113.4 million non GAAP.

The mix is improving. Products did the heavy lifting at $296.6 million with a 34.3% non GAAP margin. Services printed $54.4 million with 12.2%. Electricity sales are small at $12.8 million, but carry close to 40% margin. Installations were $37.4 million, basically breakeven on non GAAP, a bit negative on GAAP. This is the picture I want to see. Product scale. Services attach. Electricity as margin padding. Installations still the to do.

Let me quantify, because it matters for what comes next. On non GAAP gross profit, products contributed about $101.7 million, roughly 90% of the total. Services added $6.6 million, electricity $5.1 million, installations flat. If installations move from flat to 5% margin on $37.4 million of revenue, that is roughly $1.9 million more gross profit, about 45 to 50 basis points blended. If services lift from 12.2% to 20%, that is about $4.2 million more gross profit, roughly 100 basis points blended. Small lines can shift the overall picture more than people assume.

Operating leverage showed up. GAAP operating loss shrank to $3.5 million from $23.1 million last year. Non GAAP flipped to $28.6 million of operating income, about 7.1% margin. Adjusted EBITDA came in at $41.2 million versus $10.2 million last year, the kind of improvement you can actually feel running through the P and L.

That is real progress. Stock based comp still looms large at $30.2 million, about 7.5% of revenue, and it eats roughly 73% of adjusted EBITDA. So even though the optics look better, I keep an eye on quality.

Bottom line, GAAP net loss was $42.6 million, while non GAAP net income showed $22.1 million, or $0.10 per share. The reconciliation adds back $32.3 million from the debt extinguishment loss, stock comp, and a handful of smaller items.

Cash flow is no longer a fire drill. Management guided 2025 operating cash flow roughly in line with 2024 at about $92 million. A few years back this line was hundreds of millions negative. Flat to modestly positive does not make Bloom a free cash flow machine, it does remove the emergency financing label, which matters with rates where they are.

Costs still weigh on GAAP. Depreciation and amortization were $12.6 million. Interest and other were $5.5 million. Add the debt extinguishment hit, and GAAP convergence will stay slow unless the cost of capital eases or refinancing terms improve. As a check, D and A ran near 3% of revenue, interest near 1.4%, useful when you build scenarios.

Guidance is steady. Management reiterated close to 29% non GAAP gross margin and $135 to $165 million of non GAAP operating income for 2025. Q2 delivered $28.6 million. The low end implies roughly $34 million per quarter, so H2 needs to run above the Q2 pace. My simple checkpoint for Q3. Gross margin above 28%. Product at or above 34%. Services drifting toward mid teens. Installations cleanly in positive territory. Operating income with a three handle is fine, a four handle is upside.

Policy clarity around clean hydrogen credits improves project bankability. DOE hydrogen hub processes are advancing. Both are helpful for electrolyzer orders and integrated projects. Add the surge in data center power demand, where reliable, on site power and microgrid designs are getting serious attention, plus California grid reliability concerns. Bloom sits in the slipstream of multiple currents. That is the bull part of the backdrop.

The other part. If rates stay higher for longer, discount rates and financing costs remain heavy. With expectations already high, macro will act like an accelerant, or a handbrake. Which is exactly why I always check the tape.

Valuation commentary from proprietary models is out. Positioning, however, is fair game, and it is loud. The options board and hidden flow both support the recent run, and they also explain why pullbacks have been shallow.

My own model shows spot boxed between $50 support and $55 resistance. Calls dominate the board at both levels. Premiums have climbed in step with spot, while puts stayed sleepy. Traders are paying up for leverage on the way up, while skipping insurance. That is momentum money. The tell I watch for is a regime flip, a day where put premium spikes while price is flat, or only down a little. That is usually the first hint the crowd is shifting from chase into protect.

Expiration timing adds conviction. Early September looked quiet, then the September 19 and 26 contracts pulled in heavy net exposure, with cumulative load building into October and November. Deltas leaned long from call demand, not short from put buying. That profile is directional. It says traders are leaning into a multi week move, not only scalping weeklies. In practice, as long as spot lives above $50, the path of least resistance is a grind into the call wall, a pivot around $55 near OPEX, then a reset if flows stay one sided.

Open interest by strike is the clean cross check. Calls tower over puts at nearly every level, with the biggest piles at $50 and $55. When the board looks like this, the feedback loop is simple. Price up, dealers buy more to hedge. Price dips, there is not much put demand to cushion, so a break below $50 can slide quickly. If you are already long stock, this skew is your ally while price holds above $50. If you want fresh risk, I prefer scales on pullbacks toward $50, not chases through $55.

Premium flows tell you who is paying. Call premium kept climbing right along with spot. Put premium stayed dull. The moment to watch is the regime flip. If put premium surges on a flat or slightly down day, that is the first smoke of a behavior change.

Now the part I like, because it is your edge. The price action lined up almost perfectly with the buy and sell signals from my own ML model. Through the first half of the year, Bloom chopped in the mid $20s. In August, the model began flashing consistent buys right as call demand heated up. That alignment yanked the stock into the low $50s in a few weeks. The model was not just following price, it captured the underlying momentum shift as flows and positioning turned. As long as the stock stays above $50 on quiet tape, the signals remain supportive. A clean break through $55 with volume likely triggers the next cluster of buys. A failure at $55, then a close back under $50 with put premium spiking, flips the model and leaves the tape exposed.

Dark pools round out the picture. In late July and August, buy ratios flipped green and stayed that way. The big print on August 20 saw more than 6 million shares net bought, buy ratio around 2.26, DIX reading up in the 0.6 to 0.7 zone. In my playbook, DIX above 0.6 is real accumulation. The lag is usually days, not weeks. That is exactly what we saw. Price ripped soon after. When dark pools keep bidding while the options tape leans call heavy, you often get the trend that looks overextended, then keeps running anyway.

The risk to a cautious view is mostly macro. A September rate cut, then another into year end, could reset discount rates and ignite a broad melt up in growth. Softer unemployment data without real cracks would strengthen the soft landing story. In that world, flows chase, options press the upside, and momentum can carry far beyond what the current fundamentals argue for. Given how calls dominate and how dark pools have behaved, this is a real path.

Three things. One, a miss on the H2 execution bar, even small. Product margin slipping under 34%, services stalling below the mid teens, installations not flipping positive, any of those would dent the pace. Two, a regime flip in premium, where put spend spikes while price is flat. That is the crowd shifting from chase to protect. Three, a clean close below $50 with soft dark pool prints. That combination often opens air pockets because the downside is under hedged.

The tables tell me the business is healthier. The tape tells me big money believes it, at least for now. Positioning is tilted up, hedging flows are supportive, hidden buyers showed up before the breakout and still look active.

If I am already long, I keep it as long as spot stays above $50 and product margins hold up in the next print. I would consider trimming into a clean pop through $55 if premiums get frothy, then rebuy closer to a reset. If I want fresh exposure, I prefer scales near $50, not chases above $55. If we lose $50 on a closing basis with put premium suddenly spiking, I step aside and wait for the next clean setup.

Bottom line. Bloom has made real progress, and the macro tailwinds are there. The stock, however, still trades like execution will be flawless. Unless the Fed hands the market a melt up, I expect a short term correction. Not a thesis breaker, a reset that brings the risk, reward back into balance.