Understand where reality disagrees with the story, and spot the winners before it becomes obvious. The independent five-day read on the one physical constraint your deal depends on, and whether it holds on the timeline the company is raising on. Dated, verifiable, ending in commit, wait, or pass.
It scans deeptech signals across eight fields to identify the one physical constraint that decides each.
It takes one company, tests it against that constraint on its timeline, and issues a dated, verifiable prediction.
Every Read produces a dated prediction, reality grades it, and as predictions resolve they sharpen the Terminal. The Read is the engine; the Terminal is the public record built from it, and the track record is still being written, in public.
You're about to commit capital to a deeptech infrastructure deal. The science sounds right. The team has credentials. The market is large. Respected investors are already circling. But there's one question that decides whether you commit, wait, or pass, and the deck won't answer it: is the one physical constraint this deal depends on actually holding on the timeline the company is raising on? That's what the Read answers, in five days, with a dated and verifiable call.
Illustrative example. Nothing below describes a real company, deal, or client.
This is the exact structure and evidence standard of a DeepRadar Read, built on an invented deal and labeled throughout, so nothing is confidential. It is exactly the method I would run on yours.
Project Meridian, a 250 MW sovereign AI compute campus. The pitch claims 200 MW of dispatchable, majority renewable power online by Q4 2028.
Is 200 MW of dispatchable power physically deliverable to this site by Q4 2028, or is the power timeline the constraint that quietly reprices the entire investment?
The deck said "200 MW clean PPA." The underlying document was a Letter of Intent. No price, no term, no financial close. The headline power isn't contracted, and that single distinction is what reprices the deal.
The Read: the constraint does not hold on the timeline they're raising on. Not impossible. Unproven, and currently trending late. The dated, verifiable call isn't "walk away." It's: don't underwrite the 2028 power claim as derisked. Wait and reprice on a 2029 to 2030 base case, or condition the commitment, and watch the named signals to know which way it resolves.
A constraint is never called binding until the document that could prove the call wrong has been checked. That's what separates a verifiable read from a narrative summary, and it's what a deck hides.
An honest "not yet" is a valid finding. If the evidence won't support a confident call, that's the dated, verifiable result you get, not a verdict dressed up to look more certain than the evidence allows.
Optional, and most clients take it: I monitor those three signals for the next 90 days and flag the moment any of them moves. The Read stays live, because the deal doesn't stop moving the day you commit. Every Read produces a dated prediction, and reality grades it as it resolves.
"This grid-storage company claims a cost per kWh that beats lithium at scale. Is that a lab-cell result or a manufacturable cost, and which one is the deal actually priced on?"
"This company frames optical interconnect as essential to its edge. Have frontier AI operators actually committed to that architecture, or is this vendor narrative ahead of operator reality?"
"This AI native biotech says AI drove lead identification. How much is genuine AI discovery vs. a standard pipeline with AI assistance, and does the distinction change the risk?"
You're weighing a deeptech deal where a single physical constraint, power, chips, optical networking, manufacturing throughput, or energy, quietly decides whether it works, and you want an independent read on whether that constraint holds before you commit.
You want a generalist sector overview rather than the one constraint; you're already decided and just want a document that agrees with you; or you need an answer before a verifiable read can be done well.
DeepRadar is run by one person, on purpose. Every Read is researched, written, and signed by me. No junior analyst, no outsourced desk, no house position to protect. I serve as an EIC Jury Expert for the European Innovation Council, lecture on deeptech venture at ESADE and UPC, mentor founders at the Cartier Women's Initiative and Hello Tomorrow, and have taken a deeptech investment to exit. The reputation attached to each memo is mine alone, which is exactly why the read stays honest.
Bring the deal you're weighing. We scope it in a 20 minute call. If we're aligned, the 5 day clock starts and you get a dated, verifiable Read at the end. If we're not aligned, wrong question or wrong timeline, I'll tell you and point you somewhere better. No pitch, no follow up.
Tell me the deal you're weighing. I reply within 24 hours to scope a free 20 minute call.
I run a limited number of Reads each month, solo, by design.