September 17 closes, against the previous close:
| Ticker | Close | Change |
|---|---|---|
| Tempus AI (TEM) | $80.36 | +14.85% |
| Twist Bioscience (TWST) | $155.56 | +8.73% |
| Moderna (MRNA) | $158.07 | +8.55% |
| ARK Genomic Revolution (ARKG) | $51.59 | +7.82% |
| Illumina (ILMN) | $245.18 | +7.10% |
| Intellia (NTLA) | $12.26 | +6.70% |
| Natera (NTRA) | $366.31 | +6.07% |
| Eli Lilly (LLY) | $1,152.44 | +1.28% |
| S&P 500 (SPY) | $762.60 | +1.13% |
| Madrigal (MDGL) | $534.40 | +1.00% |
| Vertex (VRTX) | $516.34 | +0.60% |
The index was up 1.13%. Lilly, Madrigal and Vertex finished close enough to it to call it an ordinary day. Everything above them ran six to fifteen percent.
The line between the two groups is not size or sector. Lilly, Vertex and Madrigal sell approved drugs and book revenue against them. Tempus, Twist, Illumina, Natera and Intellia sell data, sequencing and tooling into drug discovery, and they ran five to thirteen times as far as the index.
Now the part that is easy to get wrong, and I did get wrong on the first pass: almost none of those companies had news that day.
Only Twist had a catalyst
Tempus, the biggest mover at +14.85%, had no announcement I can find dated that day. Its last two major announcements were both in July: second quarter results on July 30, showing revenue up 22% to $382.5M and a first quarterly net income of $5.6M, and the agreement to acquire Personalis at $16.25 a share, announced July 20 at roughly $1.5B enterprise value. Those are two months old. They did not move the stock on September 17.
The genuine catalyst belonged to Twist Bioscience, announced the previous day. Twist joined Lilly TuneLab as a data provider for antibody characterization, including for AbLab, TuneLab's antibody developability prediction model.
For the rest I could not find a same-day announcement either. That points at flows rather than evidence, though it is worth saying plainly that an empty news archive is weaker proof than it looks, and this piece is partly about how easily that trips people up.
What TuneLab actually is
TuneLab is a platform Eli Lilly built to give outside companies access to AI models trained on Lilly's own proprietary research data. More than 75 companies participate.
The Twist agreement works like this. A TuneLab user can order wet-lab antibody work from Twist using preferred protocols at preferred pricing, and the data that work produces goes back into TuneLab for federated training.
That loop is not the usual arms-dealer arrangement. The supplier does not sell a reagent and leave. The supplier's output feeds a model that every other participant then draws on, which makes Twist a standing input to a system that improves as more people use it. Lilly gets better models without paying for all the data itself. Twist gets designated as the preferred route for a specific kind of experiment across more than 75 companies. BigHat Biosciences holds a similar position for antibody developability datasets.
The market priced that designation as though it is worth something durable. Whether it is depends on how many of those 75 companies actually route work through it.
Sector flow and single-name news look identical for a few hours
This is the kind of session that is hard to trade on price alone.
ARKG closed up 7.82%, so a basket holder got paid for showing up. Tempus closed up 14.85%, so a name holder got paid roughly double, on no company news at all. Twist, the one name with a real catalyst, finished up 8.73%, which is barely distinguishable from the basket.
Read that back. The stock with the news underperformed the stock without it. Anyone trying to infer the catalyst from the price ranking would have picked the wrong company.
Mean-reversion strategies were structurally on the wrong side of the day. A strategy that sells strength and buys weakness trims into a 14% session by construction. Sometimes that is right, because a move on sector flow and no evidence can give itself back just as quickly. Sometimes it is wrong, because a re-rating on a partnership does not revert. The price alone does not say which one you are looking at.
That is the honest reading, and it is less satisfying than a clean story about the market rewarding data suppliers. Some of that happened. Most of the day was the sector moving as a block.
What to watch next
TuneLab has more than 75 participants and a small number of named data providers. Each new provider announcement is a potential repricing event for a company most people have not heard of, on the same logic that moved Twist. That is a watchlist, not a thesis.
The Personalis deal is still open. It is expected to close in late 2026 or early 2027, subject to Personalis shareholder approval and regulatory clearance, and shareholder firms have been advertising fairness investigations since July, which is routine for a deal this size and occasionally not. The consolidation point stands regardless: the data businesses are buying each other, and the fewer of them there are, the more pricing power the survivors hold over exactly the pharma companies that spent September 17 flat.
Twist has gone from $32.39 in the first session of the year to $155.56, up 380%. Tempus is a different shape, up 29% on the year but up 75% since the start of August. Neither of those is a forecast, and both are the kind of move that makes the next disappointment expensive.
Prices from Massive daily aggregates, September 17 close against September 16 close. Twist joins Lilly TuneLab, September 16. Tempus second quarter results, July 30. Tempus to acquire Personalis, July 20.