Nefarious Research
Est. 2021
The AI Cancer Vaccine Trade
5 min read · Aug 19 2026
MRNA143.43 +127% NTRA322.10 +21% TWST138.37 +19% TEM57.68 +17% SOPH7.71 +15% TXG60.59 +9% MRNA143.43 +127% NTRA322.10 +21% TWST138.37 +19% TEM57.68 +17% SOPH7.71 +15% TXG60.59 +9%
AI Healthcare · August 19 2026
The AI Cancer
Vaccine Trade
TEM · TWST · NTRA
+127%
MRNA on the day

Moderna's AI picked 34 targets out of a patient's tumor. It worked.

The AI Cancer Vaccine Trade | Nefarious Research
The Short Version
  • First ever Phase 3 win for an AI-designed personalized cancer vaccine. MRNA more than doubled.
  • Moderna calls it a technological breakthrough, not just a medical one. One AI picks the targets, a second runs the factory.
  • The trade is the assembly line, not the drug. Read the tumor, pick the targets, write the DNA, decide who needs it.
  • TEM, TWST, NTRA. Only TEM has a documented link. The other two are logic.
  • Everything here is already up 15-20% or more. Chasing this is risky. Not financial advice.
Read This First

Not financial advice. Research and opinion only. Do your own research and manage your own risk.

§ 01 — The Watchlist

Every AI healthcare name on my list

These are the ten I track in this basket. Nine of the ten were green on the day. This is what each one actually does.

TickerCompanyWhat it doesMove
MRNAModernaThe mRNA cancer vaccine itself+127.8%
NTRANateraSignatera personalized MRD blood test+21.4%
TWSTTwist BioscienceSynthetic DNA on silicon chips+19.2%
TEMTempus AIAI oncology, tumor sequencing plus data+16.9%
SOPHSOPHiA GeneticsHospital SaaS for genomic analysis+15.3%
WGSGeneDxRare disease exome and genome sequencing+10.8%
TXG10x GenomicsSingle-cell and spatial research tools+8.9%
SDGRSchrödingerPhysics-based small molecule design+7.3%
GHGuardant HealthLiquid biopsy, MRD, cancer screening+6.9%
LLYEli LillyBig pharma, the scale buyer of all this+3.4%
Moves are from the August 19 2026 session. GH, SOPH and WGS are marked from the earlier print in the day; the rest reflect the later tape, so the two are not perfectly synchronised. Highlighted rows are the four that matter for this trade.
Before you do anything — read this

Look at that Move column again. Most of these names are up 15 to 20 percent in a single session and Moderna more than doubled. Buying anything here means paying a price that already has the news in it. Chasing a move like this is very risky — the easy money was made by people who owned it yesterday, and gaps this size very often give a chunk back.

I am not telling you to buy any of these. This is not financial advice. Do your own research, wait for your own levels, and never chase someone else's green candle.

Why they all moved together

One event lifted the entire basket, and it was not an AI product launch. It was a cancer trial — but the therapy behind it is built by AI, per patient. That distinction is the whole piece.

§ 02 — What Happened

The first Phase 3 win for a personalized cancer vaccine

1,137
Patients
34
AI-picked targets each
First
Ever for mRNA cancer therapy

INTerpath-001 met both recurrence-free and distant metastasis-free survival in resected stage IIB-IV melanoma, randomized 2:1 against Keytruda alone. Earlier five-year data showed a 49% lower risk of recurrence or death. This is the first positive Phase 3 for any individualized neoantigen therapy, which is why the whole genomics complex moved.

ProgramCancerStatus
INTerpath-001MelanomaPhase 3 hit, Aug 19 2026
INTerpath-002Lung (NSCLC)Phase 3 enrolling
ExpansionMore tumor typesPlanned

Moderna is building a dedicated commercial manufacturing plant in Marlborough, Massachusetts. Nobody builds a bespoke factory for a therapy they expect to stay in trials. Lung is the far bigger population and the next readout that matters.

§ 03 — The Mechanism

An AI story, in Moderna's own words

Moderna's own post is titled Advancing the Fight Against Cancer through mRNA & AI, with a section headed A Testament to the Power of AI.

Moderna, on their platform
"It is also a technological one. Our vision is to build the best individualized medicine platform in the world."

"A series of fully integrated AI algorithms takes next-generation sequencing data from tumor and blood samples, reviews their genetic mutations, and predicts up to 34 of those neoantigens most likely to elicit an immune response."

It also learns — the algorithm improves by pairing clinical and immunogenicity data, so every patient treated sharpens the next selection.

Step 01
Read the tumor
Sequence tumor and blood to find that patient's unique mutations. Thousands of them.
Sequencing
Step 02 — the AI
Pick 34 targets
Predict which few will actually trigger a T-cell response. Wrong picks, useless vaccine.
Neoantigen prediction
Step 03
Write the vaccine
Synthesize mRNA encoding those targets and dose the patient.
DNA synthesis

There is a second AI nobody mentions. Moderna built Maestro, which schedules every patient batch against real-time plant capacity and re-plans automatically when anything slips, so a bespoke drug lands before that person's dose date. Designing thousands of different drugs at once is one problem. Manufacturing them on deadline is another.

In Plain English

Your cancer has a fingerprint nobody else has. One AI reads it and picks the 34 parts your immune system can be taught to hunt. A second AI runs the factory so your batch is ready on the day you need it. This does not scale like a drug. It scales like a factory.

§ 04 — The Trade

Three names, three different jobs

$TEM
Reads the tumor, picks the targets.
$61.16 · +23.9%
$TWST
Writes the DNA. Built Claude's protein designs.
$142.39 · +19%
$NTRA
Decides who needs it and if it worked.
$320.90 · +21.4%
TEMTWSTNTRA
Revenue$382.5M$118.4M$752.8M
Growth+22%+23%+37.7%
Gross margin~64%52.8%64.5%
Bottom lineProfit $5.6MLoss $35.1MEPS -$0.47
Vaccine linkDocumentedInferredInferred

TEM — the one with the receipt

The Personalis acquisition
July 20 2026: Tempus agreed to buy Personalis for ~$1.5B. Personalis' ImmunoID NeXT is the platform Moderna and Merck use for intismeran. Merck separately took a $50M stake. Personalis clinical diagnostic revenue grew 441.6% in Q2. One month later the Phase 3 read out positive.

Personalis also brings NeXT Personal, an ultra-sensitive MRD test — so Tempus owns both the profiling that designs the vaccine and a test that measures whether it worked. It is also the only one of the three posting GAAP net income, at $5.6M.

Bear read: $1.5B is a big bite, integration risk is real, and the stock repriced the same day you found out.

TWST — builds what the AI designs

SegmentRevenueGrowth
DNA Synthesis & Protein$56.6M+39%
NGS Applications$61.8M+12%

Total $118.4M, up 23%. FY guidance raised to $456-457M.

The link nobody is drawing
Twist says NGS growth came from diagnostic customers scaling tumor-informed MRD testing. That is Signatera's exact category. Twist sells panels into the same wave that lifts NTRA — two of these three are on one supply chain.
And then there is Anthropic
Anthropic tested whether Claude could design protein binders from scratch — the step that normally takes an expert weeks per target. Claude produced 1,320 designs across 15 targets, and Adaptyv Bio and Twist Bioscience independently built and tested them. 354 bound, hitting 14 of 15 targets, at a 22-35% success rate versus an industry norm of 10-15%. On target RBX1, an open contest got 9 binders from 245 candidates; Claude got 28 from 90, and its best bound ten times tighter — 3.9 nM against 45 nM.

If AI collapses drug design from months to hours, the bottleneck moves to whoever can physically synthesise and test the flood. Twist is one of two labs Anthropic used to do it.

Three caveats. The Anthropic work is drug discovery, not the cancer vaccine — positioning, not a link to intismeran. There is no verified Moderna or Merck contract, and one source pins the Aug 19 move on Twist's own guidance. And it is the weakest financially: 52.8% margin, $35.1M loss, and a $300M raise at $96 on Aug 4 against a stock now near $142. Martin Shkreli publicly called the Claude results unimpressive — worth knowing before it comes up.

NTRA — the gatekeeper

Signatera is a personalized ctDNA test built from each patient's own tumor mutations — same input as the vaccine, different job.

Why it completes the story
The vaccine is given after surgery and it is expensive. Not everyone needs it. In resected stage III melanoma — effectively the trial population — MRD positivity after resection carried a hazard ratio of 10.77 for distant metastasis and identified who benefits from adjuvant therapy. It flags recurrence about 3 months before imaging.

Signatera Genome showed 94% sensitivity and 100% specificity across five tumor types including melanoma. Commercially it is the most proven of the three: $752.8M revenue up 37.7%, oncology volume up 57.2%, full-year guidance raised to $2.85-2.91B, operating margin improved to -10.1% from -20.2%.

Bear read: still GAAP unprofitable, and no confirmed role in INTerpath. Guardant competes directly and grew faster at 44% — the melanoma-specific data is what separates them, not the growth rate.
§ 05 — The Cuts

Why the other six did not make it

NameWhy not
GHGrew faster than Natera at 44%, but that leans on colorectal screening and there is no comparable melanoma MRD data.
TXGGenuinely relevant — single-cell is how you measure T-cell response — but strip out a litigation settlement and real growth was 3%, it swung to a loss, and it is up 161% in 90 days.
SOPHRight layer, wrong size. Roughly $0.6B market cap and thin liquidity.
WGSRare disease and pediatrics. Different clinical field entirely.
SDGRPhysics-based small molecule design. Wrong modality for an mRNA vaccine.
LLYToo big for this to move the needle. It is the buyer of the theme, not a play on it.
§ 06 — The Risks

What could break this

RiskWhy it matters
Full data not outEndpoints were met. Magnitude, safety and subgroups come later. Significant is not the same as commercially meaningful.
Two links inferredOnly Tempus is documented. If the supply chain is locked with private partners or brought in-house, TWST and NTRA are a theme, not a contract.
Already repriced — the big oneMRNA doubled, NTRA +21%, TWST +19%, TEM +17% in one session. Years of adoption are already in the tape. Buying after a move this size is the riskiest part of this entire trade, and it is the reason I am not telling anyone to buy here.
Approval then payersRegulators first, then someone has to fund a bespoke per-patient therapy. Neither is guaranteed.
§ 07 — My Take

Johnny's read

Placeholder — your words go here
Tell me whether you are chasing or waiting, which of the three you want, and your levels — and I will write it in your voice.
Not Financial Advice

Research and opinion only — this is not financial advice and nothing here is a recommendation to buy anything. Every name in this piece has already run hard, several by 15 to 20 percent or more in one session, and buying after a gap like that carries real risk of an immediate drawdown. Clinical results can also fail to become approvals, revenue or share price. Do your own research, use your own levels, and size your own risk.

Want my entry levels?

Free community, live levels, every setup as it fires.
Follow @johnnylixf
Nefarious Research
AuthorJohnny Li
PublishedAugust 19, 2026
Prices are August 19, 2026 intraday unless noted; TXG and SDGR reference August 18 closes. Sources include the Merck and Moderna INTerpath-001 announcement of August 19 2026, Moderna's blog "Advancing the Fight Against Cancer through mRNA & AI" dated 19 December 2023 for the AI platform and Maestro descriptions, Anthropic's published protein binder research and Adaptyv Bio's wet-lab case study, Benzinga on the Tempus and Personalis transaction, Natera's published Signatera melanoma data, and company quarterly reports. Roles inferred from capability rather than a disclosed agreement are labelled as inferred. mRNA-4157 / intismeran autogene remains investigational and has not been determined safe or effective by the FDA.

NFA · DYOR. Not financial advice, not a solicitation, not a recommendation. Clinical and regulatory outcomes are uncertain and markets involve risk of loss including total loss of capital.