Yesterday Google announced TurboQuant: a new compression algorithm that makes LLMs (AI engine) as much as 6X smaller and 8X faster without any loss in fidelity.
Memory chip makers, which were already under pressure from the wider “Saaspocalypse”, completely shit the bed, with names like Sandisk, Micron, et al crashing instantly, those shares are continuing their spiral today…
The reaction, ostensibly, is that this will reduce demand for memory chips.
It’s funny, because I was just writing about Jeven’s Paradox in the forthcoming edition of the letter, albeit in a completely different context.
Stated simply & quickly (because I have to get on the road for a board meeting in Hamilton), Jeven’s Paradox states that when a technological breakthrough massively improves the efficiency of some thing, demand for that thing goes up, not down.
In other words, breakthroughs like this will result in more memory usage, not less – so companies cranking out RAM chips will see increased demand (provided of course, they innovate along with the space to capture the opportunity).
Not all RAM manufacturers are treated equally, I’ve put the charts of Micron (MU), Sandisk (SNDK), Western Digital (WDC) and Seagate (STX), I’ll talk about why I’m going with Micron in next week’s letter.
It certainly has the worst looking chart in terms of having fallen below it’s 50DMA and posting a solid string of red candles.
But it’s also more solid on the fundamentals than SNDK, in particular.
Micron may not be a DFV (“Deep F***ing Value”) play, but with the TurboQuant bombshell and the reaction to it, we’re flirting with maximum pessimism.
It also fits into our Post-Singularity Stack. Treat this as a supplemental for now, as I’m rethinking exactly how to fit our portfolio together going forward.
Other Trades:
- Re-entering PYPL with a small position based mainly on it being a takeover candidate, and if that turns out to be Stripe, then it maybe/could/might be the way Stripe goes public. (Highly speculative).
I did this through the Canadian TSX USD-hedged units.
- Picked up some ZONE (CleanCore Solutions) which is, believe it or not, a DOGE treasury play. Didn’t I say altcoin treasury companies are deeply flawed? Yes I did.
But this one has an interesting thesis behind it, in fact builds on a thesis we held last year when TSLA was in the portfolio (which remains back on the wishlist at the moment). More on this in the letter.
Most of this can wait until the next edition, but I wanted to put the Micron trade on while it’s still coming off hard.
The next edition should be out by next weekend.
— mark
P.S I”m deep down the Agentic AI rabbit hole and that’s been a whole journey until itself. After spending two weeks spinning my wheels trying (and failing) to get openclaw up and running, something finally clicked and now I have a veritable armada of agentic claws running, and I’ve been building an entire done-for-you platform to deploy and run these things without the pain.
easyNode.ai is almost ready to launch, if you want to get on the invite list, click here.
