I've promised this issue twice now — in #2 and again in #3. Here it finally is: the story of how the most basic number in this market turned out to be the hardest thing to get. The price.
Public to look at, not to use
Go to Cameco's investor relations page and you'll see the uranium spot price, updated and official-looking. Sprott shows similar numbers. It all looks perfectly public.
Now read the fine print. Those numbers are licensed from TradeTech and UxC — the two survey firms with the phones from #2. Cameco pays to display the number, and the terms forbid anyone from scraping it, republishing it, or building anything on top of it. Sprott's terms say the same.
So you can look at the price. You just can't use it.
What the license actually costs
Fine, I thought. How much can a number cost?
Thousands of dollars a year. The professional data services from TradeTech and UxC are priced for utilities, producers, and funds — people for whom the fee is a rounding error on the deals it informs. For a free newsletter and a free dashboard built on evenings and weekends, that's a non-starter.
And here's what surprised me more: even paying wouldn't solve it. Most of these licenses buy the right to see the data, not the right to republish it. Putting the number on a public dashboard typically needs a separate, more expensive redistribution agreement.
Facts are free, feeds are not
After a lot of reading (usual caveat: I'm a student building a dashboard, not a lawyer), one distinction turned out to matter most.
A data feed — the weekly spot price series, updated and machine-readable — is a product. It's owned, licensed, and protected.
A fact reported in the news is different. When TradeTech puts a price in its own press release, or a news outlet reports where uranium traded last month, that number becomes a publicly reported fact. Facts can be cited, with attribution, the way any journalist or researcher cites them.
That distinction is narrow, but it's everything. The live weekly number is out of reach — but a trail of monthly public reference points exists, scattered across press releases and news archives. I collected 52 months of them by hand.
Why nobody fixes this
My first reaction was mild outrage: it's 2026, why is a commodity price locked in a vault?
But sit with it and the logic holds. There is no uranium exchange — every deal is a private phone call. A "market price" only exists because two firms call everyone, week after week, and triangulate what they hear. That work costs money. The number is their product.
And nobody in the industry has an incentive to change it. Utilities and producers can afford the license. The survey firms live off it. Which leaves everyone else — retail investors, researchers, people building tools — outside the vault.
Three ways out
If you want to build anything with uranium prices, you have exactly three options.
Pay. Thousands a year, and the redistribution problem remains. Doesn't work for a free, open project.
Scrape. Take the number from Cameco's or Sprott's site and hope nobody notices. Plenty of projects quietly do this. I won't — it violates the terms in black and white, and a project about doing things honestly in public can't be built on a quietly dishonest foundation.
Estimate. Don't republish the number. Model it. Because while the uranium price itself is locked up, things that track it trade in plain sight: the Sprott Physical Uranium Trust and Yellow Cake PLC hold nothing but physical uranium, and their shares trade on real exchanges that anyone can query freely and legally. So you take the public numbers, let a model learn the relationship, and calibrate it against those monthly public facts from the news trail.
That's the QL Spot Estimate. That's why it exists, why it's called an estimate, and why it's checked against reality every month in the open. The maze is the reason it has to exist.
The takeaway
When a website shows you "the uranium price," someone paid to display it, and it legally can't flow any further. The market's most basic number is a licensed product, not a public good.
How well a model can actually stand in for the real thing — error margins, what it catches, where it fails — that deep-dive is coming. Some of the numbers surprised me.
Also: some news about this project next Sunday. Stay tuned.
— Maximilian