I keep hearing the same line: “data centers will figure out power.”

Maybe. But in the near term, it usually looks like a scramble. Interconnection delays. Upgrade schedules that slip. Peak demand that shows up right when you do not want it.

So I went back through Fluence Energy’s Q2 FY2026 release from May 6, 2026. Not to romanticize it. Just to see what was actually said, and what was actually booked.

Numbers first, because otherwise it turns into vibes.

Revenue in Q2 FY26: $464.9 million

Adjusted gross margin: 11.1% (GAAP gross margin: 10.0%)

Contracted backlog as of March 31, 2026: about $5.6 billion (record)

FY26 guidance reaffirmed: revenue $3.2 to $3.6 billion, adjusted EBITDA $40 to $60 million

At first glance, that revenue line is the “problem.” It landed well below what a lot of people were expecting. Management blamed port and shipping delays.

Here’s the thing. I do not think you can dismiss that explanation, but you also cannot just wave it through. Logistics issues have a way of coming back twice, especially when the whole industry is trying to move heavy equipment at the same time.

Still, the backlog number is doing the heavy lifting. $5.6 billion is not an “interesting” backlog. It is a backlog that can force the market to pay attention, even if the quarter itself was messy.

Then there was the line that made people sit up: Fluence said it signed master supply agreements with two major hyperscalers, and it expects the first order during fiscal Q3.

They also described a data center pipeline around 12 GWh, and said the large majority is tied to those two supply agreements.

What’s interesting is how that changes the tone of the business. Utilities can be slow. Data center buyers can be slow too, but once they are in motion, they do not like stopping. They hate stopping.

Slight tangent, but it matters: people talk about “new generation” like it is a switch you flip. In reality, bridging solutions get a lot of the work done while everyone waits for the big projects. Storage is one of the few levers you can pull in pieces, on timelines that are sometimes measured in months, not years.

There is also the product angle. Fluence said it reached substantial completion on the first Smartstack delivery in the quarter. I like seeing that, because new offerings tend to be painful before they are helpful. Either way, you get information quickly.

My hesitation is still margins. Adjusted gross margin was 11.1% in Q2 FY26. To land full-year adjusted EBITDA of $40 to $60 million, the back half of the year has to be better. Cleaner deliveries, fewer surprises, fewer “one-time” issues that keep showing up.

So here’s where I’m at.

If that first hyperscaler order actually arrives in fiscal Q3, and it looks like the beginning of a program (not a toe in the water), it changes how you have to model the next couple of quarters.

If it does not arrive, or if it arrives tiny, the market’s patience gets shorter fast.

Take a closer look.