21 Comments
User's avatar
Dr. K's avatar

This is a more brilliant piece of work than the credit you will receive. Superb analysis and likely to be mostly correct. The still unsolved part of the problem is the understanding engine for the phenomics. This has proven to be far harder than imagined (every patient is their own science experiment) and is the source of much feverish activity in the cognitive AI space. It is still the missing piece to pulling off your thesis. But as close as I have seen to future predicting in quite a while.

Expand full comment
Charlie Cauldron's avatar

I think what you're describing -- making bets on which companies will succeed over time -- is called having a stock portfolio. If you're this good at it )and it seems like you are) instead of making bets with friends, buy shares in the company next time.

Expand full comment
Francis Turner's avatar

Frequently though promising companies are privately held.

Expand full comment
Marton's avatar

Not this time.

Expand full comment
Francis Turner's avatar

Nothing stopped our host from investing as well in that case. The fact that he didn't mention that doesn't mean he didn't do it

Expand full comment
Brendan Long's avatar

Investing is much more complicated than making a single well-specified bet though. Lots of people are correct about things and fail to come up with the correct trade to profit from it, or lose money because of unrelated events changing prices.

Expand full comment
Sol Hando's avatar

Looking at their balance sheet, 23andMe had a LOT of problems, including high cost of revenue, (exceeding their revenue when you include admin and marketing) and very high R&D expenses. It seems like they had a fundamentally good business model, selling super high margin DNA testing kits (these things should cost less than $10 not including shipping), but it's been mismanaged to hell as far as cost cutting.

Somehow administrative and general expenses were HIGHER than their cost of revenue, which signals absolutely terrible bloat.

I have no idea what they do with that R&D money, but an acquisition by a responsible corporation (especially one capable of doing its own R&D already and with its own admin team) should be able to bring them to profitability relatively quickly, and the cost of acquisition was not much more than the book value of the company.

I have no idea why they acquired lemonade Telehealth at a time when they were already burning money. Clearly tele-health is a profitable business model, and they have a beautiful lead funnel through their genome testing kits ("You're 20% more likely to have depression, want to talk to a therapist about it?"), but why spend more than the book value of your company to acquire Lemonaid Health? Telehealth is a capital light business model, and they could probably have built their own team in-house, scaling it to the demand that 23andMe was able to generate, without having to acquire a whole business at a premium.

Expand full comment
Realist's avatar

Thanks for the great article. The wedding of DNA research and pharmaceutical development is the future.

Expand full comment
Steve Hemingway's avatar

Terrific analysis.

Expand full comment
L N's avatar

Any opinion about the time horizon where we will start seeing if Regeneron’s bet is paying off?

Expand full comment
barnabus's avatar

23andme is still cheap compared to Regeneron - I think they will see an impact within 5 years or so. The problem for other pharmaceuticals is that there are only 2 comparable companies around with that size, ancestry and myheritage.

Expand full comment
SurgeStick's avatar

Any estimate on what percentage of clinical-trial cost could be saved by cutting overly burdensome regulations and/or how much disutility these regulations cause in aggregate?

Expand full comment
Cremieux's avatar

Depends on the modality, but the amount is uniformly substantial.

Fixing GMP alone could save some drugs around 50%!

Expand full comment
ardavei's avatar

Why just 50%? The price differential between research use only (RUO) and GMP-compliant compounds for research use is often more than 10x.

Expand full comment
Christoph's avatar

Great post... I wonder what you make of the other strand of promising research that could de-disk pharma R&D, done by companies like Scipher, whose proposition is based on the work of Laszlo Barabasi on the protein connectome. Has been clinically successful and also got Medicaid approval etc..

https://www.sciphermedicine.com/wp-content/uploads/2021/02/Cheng-Kovács-Barabási-2019-Network-based-prediction-of-drug-combinations.pdf

Expand full comment
Christoph's avatar

Links broken, here's a better one: https://www.nature.com/articles/s41467-019-09186-x

Expand full comment
Mike's avatar

I have a question – was there any bidding war between drug companies? Why Regeneron and not Bayer, Novartis etc?

Expand full comment
barnabus's avatar

I would guess there may have been an overlap through private equity. IE same partners holding Google, 23andme (a kind of Google baby) and Regeneron. Bayer and Novartis are 125y+ dinosaurs that don't have that big private equity investorship. Bayer also "lost" a lot of private equity investors after 1933...

Expand full comment
barnabus's avatar

Sincee Regeneron deal came up on the horizon, 23andme exploded from the lows of 0.64$ to something like $3.40, over the last 5 days or so. It is now selling at prices not seen since last year...

Expand full comment
ardavei's avatar

The actions of the current administration regarding academic biomedical research would also tend to favor Regeneron. With the current grant chaos and funding cuts hitting academic labs, the advantage of having in-house data becomes stronger.

Of course this will slow down the general rate of biomedical progress, but it will advantage Regeneron specifically.

Expand full comment
barnabus's avatar

Let's not forget that Regeneron provided Donald with experimental antibody when he was down with Wuhan Corona virus :D

Expand full comment