Reverse Product Discovery on SPAN's residential data center product, XFRA

👋🏼 Hi, I’m Moiz. I'm a Fractional Head of Product who helps top climate-tech teams build product from strategy, discovery, to execution. You’re receiving this because you’re building in climate and we’ve crossed paths. Each issue distills one proven concept that I use with my clients - and gives you a way to apply it to your work.

Want to get sharper at product discovery?

Reverse engineer a great product from a company and people you admire. Understand the underlying assumptions of a product that was launched.

You'll quickly see the risks they took and why the company decided to Build, instead of Pivot.

I call this Product Discovery, Reversed. Here's how it works:

1. Pick a company and product you admire.
2. Then, imagine you're the PM responsible for this product.
3. Define the assumptions the product team made
4. Identify the riskiest assumptions and ask yourself: Build or Pivot?

Do this enough and you'll start to see the hidden assumptions in your own products. And you’ll learn how to de-risk your own bets so you can build a successful product.

I ran that exercise on SPAN, the company turning homes into tiny data centers. With the help of VERA, the AI I built for customer discovery, I clearly saw what it would take for this product to succeed and why it might fail.

Telling me why my Idea will fail is my killer AI use case in product.

Here's the TL;DR:

1. SPAN is launching XFRA, a tiny inference datacenter sited at residences, to help hyperscalers meet their compute needs. Homeowners get the tiny data center + SPAN 200A smart panel + home battery + $150/mo subsidy all for $0 down.
2. This solves the problem of hyperscalers facing 3-5 year delays due power infrastructure constraints. Residential and commercial buildings have unused electrical capacity that could serve data centers, with no grid infrastructure upgrades.
3. The go to market is new construction homes and “build to rent” homes, in partnership with Pulte Group, a residential construction company. This is smart because they don’t have to sell existing homeowners.
4. Risky Assumption #1 - Compute Offtakers will actually purchase compute from distributed, residential data centers, at a price that sustains SPAN. Classic Product Market fit risk.
5. Risky Assumption # 2 - The second Riskiest Assumption is that SPAN’s hardware and software can meet enterprise data center requirements (performance, security, uptime reliability, operational safety). Technology Risk that Silicon Valley companies are expected to take.

SPAN is taking a pilot approach to de-risk the product and business. Shoutout to SPAN for calling out the biggest risks to the product in their white paper. That’s the clearest signal to me that this company can figure it out.

Be the PM who ships a great product. Sign up now

​

The hidden risks of launching residential data centers

The first thing VERA does is ask you to input the Idea. Here’s what I gave her:

SPAN puts small AI data centers ("Nodes") inside people's homes — for free — using the spare electrical capacity most homes never touch. It runs AI workloads on them and sells that computing power to big AI companies, while paying the homeowner ~$150/mo toward their energy and internet. The bet: stand up AI compute in months by tapping homes, instead of taking years to build a giant data center.

This Idea, like all product and business Ideas, is built on a set of Assumptions that must be true. So VERA asks you to identify the Assumptions you’re making about your Idea. And then it challenges your Assumptions by telling you it might fail. Finally, you go back and score each Assumption with High, Medium, Low risk.

I call this the Validation Stack. Here's what VERA and I generated for SPAN's XFRA (click to enlarge):

SPAN said Build, not Pivot for XFRA

The Riskiest Assumptions are whether a Compute Offtaker will sign a contract with SPAN and whether SPAN's technology can deliver.

Obviously, SPAN landed on Build for XFRA.

It’s a compelling offering and they are betting on AI growth. From the outside it looks like a no-brainer … but if it does fail, I wouldn’t be surprised if it was due to one of the assumptions above.

Every product fails because one of the underlying assumptions did not hold up to reality.

Don’t let this happen to you.

Running a discovery sprint will probably force you to kill more ideas than you Build. That’s a strength.

Try This

Use my free Claude Skill to run an Assumption test for an Idea you're building. In less than 10 minutes, you'll have more clarity on your product and know the riskiest assumption that could kill it.

​Try it for free.​

You’ll become a much sharper Product Manager or Founder if you can de-risk your assumptions upfront and then launch your product.

VERA makes it dead simple for any builder to know if their idea is worth building or not, within minutes.

From Insight to Action

• Want to be the PM who knows what to build? Join my free workshop and I'll show you how to use Claude to do it.
• Looking for your next job? Curious about trends in climate and product? Join my next Lean Coffee for Climate Product Leaders. Free, in partnership with My Climate Journey.