AN INVESTMENT THESIS · MICHAEL BLOCH · QUIET CAPITAL · APRIL 2026
Biosecurity.
The cybersecurity industry is worth over $200 billion. Biosecurity — the industry that defends against biological threats, natural or engineered — is almost nothing. That gap made sense when designing a pathogen required a state-level lab. It doesn't anymore. AI, CRISPR, and commercial DNA synthesis have made biology a programmable medium over the past three years, and the defensive perimeter has barely moved.
This page is my attempt to map the field, plainly. Inside: the 21‑step lifecycle of a biological attack (with my read on which steps are actually investable), five threat sources any durable company must work against, five cautionary tales from the last generation of biosecurity startups that died between crises, the five categories I'd back today, the seven companies I'm actively watching, and four ways this thesis could be wrong.
Scroll for the full argument, or use the nav at the top to jump. If you disagree with something — please ping me.
$200BCybersecurity Market
~$0Biosecurity Equivalent
10,000×Offense / Defense Cost Ratio
0.43%Of US Defense Budget
§ IIThe Asymmetry01 / 09
On cost, coverage, and attention
A poxvirus costs one hundred thousand dollars. The response costs billions.
In October 2025, researchers at Twist Bioscience and Microsoft showed that
artificial intelligence could design variants of known toxins that evade
every piece of screening software currently in use[1]. The paper was not a
warning about what might happen. It was a demonstration of what had already
become possible. This thesis begins there — at the moment the expertise
barrier collapsed and the defensive perimeter did not move.
Offense · per event
$100K
Estimated cost of synthesizing a dangerous pathogen using commercially
available tools and published literature.[2]
Source · O. Scharfman, 2025
Defense · per response
$13.8T
RAND's estimate of global economic losses from COVID-19. The ratio
between an attack and its consequences is unlike any other category
in national security.[3]
Source · RAND Corporation
§ IIIThreat Sources02 / 09
What we are defending against — and why the list matters now
Biosecurity is not one problem. It is five — and all five just got cheaper.
The threat sources themselves are not new; they are roughly the same ones
biodefense has tracked for fifty years. What changed is the price of entry.
AI-assisted design, commercial DNA synthesis, and CRISPR have compressed the
barrier so far that work requiring a state-level program in 1990 is
within reach of a determined individual today. A durable biosecurity company
has to work against several of the sources below — single-origin bets
("anti-bioweapons" alone, "pandemic prediction" alone) are the ones that
historically die between crises.
Source 01
State programs
2025 State Dept report: Russia and North Korea maintain active offensive programs. PLA identified biology as "a domain of military struggle" in public 2017 textbooks. Low probability — maximum severity.
Low P · Max severity
Source 02
Accidental release
Lab escapes, failed biosafety, mis-handled culture collections. Historical base rate is higher than policy discourse suggests. EO 14292's gain-of-function pause reduces — does not eliminate — this surface.
Known base rate
Source 03
Natural emergence
Zoonotic spillover, antibiotic resistance, novel vector-borne disease. The most frequent source historically. Climate and habitat encroachment are widening the surface.
Most frequent
Source 04
Agricultural
Livestock disease, crop pathogens, food-supply contamination. Agriculture is 30.2% of the biosecurity market — and almost nobody in venture is looking. The category with the clearest commercial buyer today.
30.2% of market
Source 05
Information warfare
Disinformation about outbreaks, adversary-seeded panic, weaponized uncertainty. Not a pathogen problem but a biosecurity one — response quality is gated on public trust in data.
Emerging
What changed
The barrier collapsed.
Two changes in the last three years convert each source above from "rare and expensive" to "accessible." (1) Frontier LLMs answer the how-to questions that used to require a PhD. (2) Commercial DNA synthesis + benchtop synthesizers + cloud labs shrink the physical lab from a building to a credit card. The implication: lone actors now appear on threat lists that used to be populated only by states.
The new variable
§ IVThe Full Lifecycle03 / 09
How this was built
Twenty-one steps from intent to recovery.
To figure out where durable companies could actually be built, I wrote a
short fictional version of what a biological attack would look like end-to-end —
from the moment someone decides to cause harm all the way through long-term
hardening after an event. That narrative turns into twenty-one discrete
steps across four acts.
At each step I mapped what an attacker would be doing, what a
defender would need to do to stop or detect them, what already exists,
where the gap is, and whether there is a venture-scale business at that layer.
Click any card for the full breakdown — the offense and defense view side by side,
the state of the art today, and the key question I would ask a founder
operating there.
Investability Lens
“
Split-order attacks can circumvent synthesis screening entirely by
distributing DNA fragments across different providers — none of whom
ever sees a complete dangerous sequence.
— RAND corporation, cloud-lab surface analysis
§ VCautionary Tales04 / 09
Pattern of the last generation
Every previous biosecurity company died between crises.
Post-event procurement surges, budgets normalize, gov-only business models
run out of oxygen. Commercial revenue is not a bonus — it is survival.
Metabiota
2008 — 2022 · FOUNDED · ACQUIRED
FOUNDEDABSORBED
$40M raised over a decade. Pandemic risk modeling for insurers and governments. Absorbed by Ginkgo in 2022 during the post-COVID reshuffle. Technology was strong; business model was gov-dependent.
$40M → acquihire
Sherlock Biosciences
2019 — 2024 · BROAD SPIN-OUT · SALE
LAUNCHSALE
CRISPR-based diagnostics, a Broad Institute spin-out. First COVID molecular test authorized under EUA. Sold assets in 2024 for a small fraction of capital raised. Science was real; demand was cyclical.
$5M exit
Ginkgo Biosecurity
2021 — 2025 · SCALED · DIVESTING
PEAKDECLINE
Built on the COVID contract surge. Revenue down 30% YoY to $37.4M. Being divested from Ginkgo's core business. Case study in procurement cliff — the most recent and cleanest example.
-30% YoY
§ VIConviction Ranking05 / 09
Where I would bet, in order
Five categories, ranked by conviction.
Ordered by a mix of commercial viability, technical defensibility, and proximity to regulatory tailwinds. These are not predictions. They are the priors against which I evaluate inbound.
i.
Environmental & Wastewater Metagenomics
Clearest near-term path. UK mSCAPE is proof at national scale. Commercial buyers exist — airports, schools, transit, building operators. Wastewater may capture ~80% of the value cheaply.
ii.
AI Pathogen Characterization & Countermeasure Design
Transformative if it works at crisis speed. No AI-designed therapeutic has been validated under real pressure yet. Watch: Valthos, Red Queen Bio, SeqDesign.
iii.
Agricultural Biosecurity
30.2% of the biosecurity market in 2024 and almost nobody in venture is looking. Real commercial buyers today. Returns likely modest but durable.
iv.
DNA Synthesis Screening
EO 14292 mandates screening; OSTP tightens the window to 50 nucleotides by October 2026. SecureDNA is nonprofit. Timing-dependent, with the chokepoint itself slowly eroding.
v.
Cloud-Lab Biosecurity
RAND flagged zero screening across cloud labs in November 2024. Gap is real. Business model waits on regulation naming the buyer.
§ VIICompanies I'm Watching06 / 09
Active diligence, not recommendations
Eight companies, five categories, one nonprofit.
A public snapshot of who I am spending time with today. Some I've met,
some I've written to, some I am still reading. Inclusion is not endorsement —
this is the working list behind the conviction ranking above.
Pilgrim
Pre-seed · $4.3M
Surveillance · portable · multi-threat
ARGUS Scout: a 35-pound portable automated biosurveillance platform — air collection, sample prep, qPCR against eight known bioagents, untargeted sequencing via Oxford MinION, and onboard bioinformatics in one box. Detects known, novel, and engineered threats. Three form factors planned. Vision: a "biological NORAD."
Real-time pathogen detection paired with AI-driven adaptation of existing medicines. Positioned at the highest-impact steps of the response — characterization and countermeasure design — with a dual-use pharma + biodefense thesis.
Founder · Tess Van Stecklenburg (ex-Lux) · Investors · OpenAI, Lux, Founders Fund
Red Queen Bio
Seed · $15M
Red-team · threat-mapping · AI-native
AI-driven biological threat mapping and countermeasure design. Takes offense seriously — red-teams the frontier of what bad actors can do, then works the defense side. Narrower market than Valthos but more differentiated intellectual position.
Founders · Hannu Rajaniemi, Nikolai Eroshenko · Investors · OpenAI, Cerberus, Fifty Years, Halcyon
SeqDesign
Stage undisclosed
Protein design · binders · antivirals
AI-driven protein design — binders, antivirals, enzymes. Strong dual-use implications: the same tools that design therapeutics could design threats. The technology is real; the GTM and guardrails are the open questions.
Commercial DNA-synthesis screening platform — screens synthetic DNA/RNA orders against pathogen sequence databases and runs KYC-style customer screening for synthesis providers and benchtop-synthesizer manufacturers. Directly answers the open question I raise in Step 04 of the lifecycle: if SecureDNA becomes the nonprofit default, Aclid is the commercial layer selling screening as a service to providers and equipment-makers.
Founders · Kevin Flyangolts, Matthew McKnight (ex-Ginkgo biosecurity) · Investors · AIX Ventures, SV Angel, Cantos, Schmidt Futures
SecureDNA
Nonprofit · Standard-setter
Synthesis screening · universal standard
Building the universal screening layer for DNA synthesis orders, positioned to become the default under EO 14292. Nonprofit structure is a feature (neutrality, adoption) and a question (does it leave room for a commercial competitor?).
Model · Nonprofit standard · Related to · OSTP mandate, Oct 2026
Metagenomic sequencing of environmental samples (foot swabs, waste streams) inside poultry barns to build per-facility "microbiome fingerprints" and flag pathogens before clinical symptoms appear. Founders ported the COVID-era wastewater-surveillance playbook to livestock. The clearest agricultural bet on my list and the one most aligned with all-threat-source thinking — zoonotic spillover (H5N1) is where the next human pandemic likely starts.
Investors · Twelve Below, Max Ventures, Banter, Planeteer · Buyers · Vital Farms, West Liberty Foods, Mississippi State
Ginkgo / Tower Biosecurity
Public · divesting
Deployed infrastructure · potential carve-out
Revenue down 30% YoY to $37.4M, being divested from the Ginkgo parent. Carries real deployed surveillance infrastructure and COVID-era customer relationships. Potentially the cleanest acquisition target in the space — if there is a buyer with a durable commercial thesis.
Status · Carve-out pending · Watch for · Acquirer identity
§ VIIIWhy Now — The Record07 / 09
The eighteen months that shaped this thesis
Eighteen months, eight events.
Most of what you just read — the threats, the framework, the companies,
the cautionary pattern — has been shaped by a short sequence of things
that actually happened between late 2024 and today. These are the eight
moments that matter most. If any of the eight below get reversed or
falsified, the thesis shifts.
Nov 2024
RAND: Cloud labs have zero biosecurity screening
RAND Corporation publishes a report documenting that the attack surface spans three uncoordinated systems — AI design tools, commercial synthesis providers, and cloud laboratories — with no consistent oversight. Cloud labs, which can run biology experiments for remote customers, perform no pathogen screening whatsoever.
RAND Corporation · public report
Jan 2025
UK mSCAPE goes live — first national metagenomic biosurveillanceMetagenomic biosurveillanceSequencing every piece of genetic material in a sample — not just what you were looking for. Detects known pathogens AND organisms never seen before. The replacement for legacy "known-pathogen-only" systems like BioWatch.
The UK Health Security Agency begins operating mSCAPE, the world's first national program to sequence unknown pathogens from clinical samples. Proof point that metagenomic surveillance can work at country scale.
UKHSA · operational
Apr 2025
NSCEB final report: $15B federal investment over five years
The National Security Commission on Emerging Biotechnology publishes its final recommendation — $15B over five years, a national biosurveillance backbone, and dedicated manufacturing resilience. Congressional appropriation remains an open question.
NSCEB · congressional commission
May 2025
Executive Order 14292
Pauses gain-of-functionGain-of-functionResearch that enhances a pathogen's transmissibility or lethality to study it. Controversial for years. The EO pauses it — but the underlying tools (synthesis, AI design, CRISPR) are not paused. research, mandates universal synthesis screeningSynthesis screeningWhen you order DNA from a commercial provider, software checks it against known dangerous sequences. The current US standard is a 200-nucleotide window; EO 14292 tightens it to 50 nucleotides by October 2026., directs OSTP to tighten the screening window from 200 to 50 nucleotides by October 2026. First time synthesis screening moves from voluntary to required.
White House · executive order
Oct 2025
Twist / Microsoft · Science
Researchers demonstrate that AI-designed protein variants of known toxins evade every major commercial screening algorithm. Not a prediction — a published capability. This is the single most important paper in the field.
Science · October 2025
FY 2026
US biodefense spending falls — $3B cut
FY2026 biodefense appropriation: $5.12B, down from $8.16B in FY2025. CDC halved. NIH $40B → $27.9B. The COVID-era contract surge unwinds. The offense moves forward as the defense pulls back — the central paradox of this thesis.
Congressional Budget Justifications
Early 2026
State Dept adherence report confirms active offensive programs
The annual State Department compliance report affirms Russia and North Korea continue to maintain active offensive biological weapons programs, in violation of the Biological Weapons Convention. PLA public doctrine since 2017 treats biology as a domain of military competition.
US Department of State
Oct 2026
OSTP screening-window deadline
Under EO 14292, DNA synthesis providers must screen orders against a 50-nucleotide window — a material tightening. This is the concrete forcing function creating a commercial screening market. The closest thing to a near-certain catalyst in biosecurity.
Office of Science and Technology Policy
§ IXWhat Would Make This Wrong08 / 09
Disconfirmation, in advance
The ways I could be most wrong.
Publishing these now so I can be held to them later. If any of these turn out to be true, the thesis collapses or changes shape.
Falsifier 01
The AI bio-threat is overhyped.
CSIS (Aug 2025) and the National Academies argue current models cannot design novel self-replicating pathogens. If the capability gap stays wide, urgency dissolves.
Falsifier 02
Government handles it — quietly and competently.
BARDABARDAThe Biomedical Advanced Research and Development Authority. The federal agency that funds and procures medical countermeasures — vaccines, therapeutics, diagnostics. Post-9/11 creation; the closest thing the US has to a durable biodefense customer. has spent over $80B since 9/11 and produced real countermeasures. If the federal stack is sufficient, private capital has less to do here than I think.
Falsifier 03
Spending is moving the wrong way.
FY2026 biodefense: $5.12B, down from $8.16B. CDC halved. NIH $40B → $27.9B. The market may be shrinking faster than it is forming.
Falsifier 04
The catalyzing event never comes — or comes too late.
Sustained procurement has historically required a triggering event. Pre-event companies may not survive the interval, no matter how good their technology is.
Primary reading, in order of how much it shaped my thinking
Sources & methods.
Every numbered citation on this page maps here. If you want to push on
the thesis, these are where I would start.
O. H. Scharfman — Notes on US Biodefense Spending.
Substack · 2025 · 0.43% of defense budget analysis
[9]
CSIS & National Academies — Assessments of AI / biosecurity threat models.
CSIS · August 2025 · capability gap critique
[10]
US Department of State — Annual adherence report to the Biological Weapons Convention.
State Dept · Russia / North Korea · 2025
— END OF DOCUMENT, SECTION ZERO —
The defensive perimeter is not moving on its own.
I'm building this thesis in public. If I missed something, if you have data
that contradicts me, or if you disagree with a specific step, ranking, or company
take — please ping me. I'd rather be corrected now than confident and
wrong later. Also reach out if you're building in the space, or just want to talk.