20VC: $5BN in Revenue, 7 to 7,000 Employees in 9 Months, 206,000 Tests in a Single Day: The Craziest Story in Startups: Curative with Fred Turner

Summary

Fred Turner recounts how Curative moved from earlier diagnostics work into pandemic testing after its chief scientific officer built a COVID test in spare time just as lockdowns began. The company hacked around lab licensing, built an “orthogonal supply chain,” and rapidly scaled to 206,000 tests a day, 7,000 employees, and roughly $5 billion in revenue over three years. Turner says the economics were brutally cyclical: surge periods were profitable, but dips were loss-making because the infrastructure had to stay in place, and even vaccination was a money-losing business. Curative later pivoted toward health insurance, which Turner argues is a better fit for AI-driven automation because the payer drives behavior in U.S. healthcare and admin work is still bloated. He makes a strong case that current models can already run Curative’s back office, while the deeper challenge is deployment, exception handling, and trust in agents acting on behalf of humans.

Notes

• Curative began with a COVID test built in spare time by its chief scientific officer before lockdowns began.
• The company was originally positioned to pivot from earlier diagnostics work into pandemic response.
• Curative’s COVID business ultimately produced about $5 billion in revenue over three years.
• Peak daily testing reached 206,000 people.
• Headcount went from roughly 7 to 7,000 in nine months.
• Curative also administered 2.5 million vaccinations, but Turner says it lost money on that service.
• The company is cutting about 80% of SaaS spend.
• Turner says Curative’s earlier company-building path was influenced by Silicon Valley’s willingness to back younger founders.
• He contrasts the UK funding market with Silicon Valley’s founder-friendly ecosystem.
• Curative’s first product work started in cattle genomics, predicting traits like musculature and milk yield.
• The cattle business had a built-in biological tradeoff: too much musculature can cause calving problems.
• Turner says a last-ditch fundraising attempt pushed him toward Y Combinator and the U.S. market.
• The cattle-testing company raised a seed round and later hit a TAM ceiling problem.
• The cattle-testing opportunity was too small to support a large venture-scale company.
• Curative shifted from agricultural genomics into human diagnostics because the market was larger and paid better.
• The company looked for diagnostics areas where DNA testing could create a stronger treatment decision.
• Turner frames sepsis as a high-mortality condition driven by the body’s own immune response.
• He says every hour of delay in sepsis treatment raises mortality by about 12%.
• The early sepsis product aimed to detect bacteria and antibiotic susceptibility quickly.
• Turner says sepsis tests looked good in smaller settings but failed to replicate in larger studies.
• The original Curative sepsis pitch was a fixed-fee “mini hospital in a hospital.”
• Curative’s first outside investor was Justin Mateen, who put in $125,000 at a $3 million valuation.
• The initial hospital pilot in Wisconsin stalled because the hospital was preparing for COVID.
• That delay was Turner’s first real signal that COVID would become a major event.
• Curative’s COVID test was initially meant as an internal training exercise.
• Public demand was initially low because nobody wanted COVID-related work onsite.
• Curative formed a 50/50 JV with a lab-license holder before buying it out.
• Curative paid $27 million to buy a lab license after previously selling one for $150,000.
• The first real government customer was the San Dimas sheriff’s department.
• Curative emphasized sample deactivation so there was no live virus on site.
• The City of Los Angeles contract came from a tweet that turned into a DM conversation.
• Los Angeles paid after delivery on net-one terms, and Curative sent someone to city hall every morning to pick up a check.
• A Florida nursing-home testing contract was worth hundreds of millions of dollars.
• Florida tested every nursing-home employee weekly for three months.
• Turner says incumbent labs are built for efficiency, not 10x scale.
• He says the big labs are optimized for incremental margin improvements.
• Curative deliberately built an “orthogonal supply chain” instead of using standard lab tooling.
• Curative sourced swabs from electronics vendors and sterilized them.
• The company replaced magnetic beads with filter plates made of glass and plastic to improve scalability.
• Curative hit 206,000 tests in a single day in December 2020.
• The scale-up from zero to peak testing volume happened in about eight months.
• Drive-through sites were run in parking lots rather than conventional facilities.
• The Dodger Stadium site had seven lanes and operated 7am to 7pm, seven days a week.
• The Dodger Stadium site tested about 10,000 people a day at peak.
• Testing economics were good during surges and bad during lulls because fixed infrastructure had to stay up.
• Turner says Curative lost money on every test during the dips.
• Government reimbursement was raised to incentivize peak testing capacity.
• Curative’s vaccination service lost money because reimbursement was below administration cost.
• Turner says the company initially told new hires the role would only last three months.
• Curative began taking health insurance seriously in late 2021.
• Turner says lab testing alone was too small to build a giant company.
• He rejected hospitals as a pivot because payer mix is fragmented and hard to control.
• Turner argues the payer, not the provider, drives behavior in U.S. healthcare.
• Independent primary care doctors get paid about half as much as system-affiliated doctors for the same service.
• Turner says the U.S. payer market is highly concentrated.
• He says the U.S. has the most advanced facilities and drugs and spends more to preserve life.
• Turner wishes he had understood the AI wave when designing Curative’s insurance business in 2022.
• He describes health insurance as a business that mainly moves bits in databases.
• Curative built an in-house Claude agent to automate credentialing end to end.
• Credentialing time dropped to about 12 hours and cost to about 20 cents.
• Claims processing is also being automated with agent-written Python transformations.
• Curative canceled a $600,000 annual Salesforce contract after building an internal CRM.
• The company is cutting about 80% of SaaS spend this year.
• Curative built its own claims system from scratch and expects to be off the legacy system by July.
• Turner thinks some of the largest insurers will struggle to match agent-driven margin gains.
• He notes that 85% of premium must be spent on care.
• That rule caps insurer profit at about 15% and pushes incentives toward higher spending.
• Anthropic spending at Curative has grown about 6x every month for six or seven months.
• Curative’s network contracting problem covers about 1.2 million providers and 60,000 to 70,000 contracts.
• A 45-person team previously handled provider network contracts.
• Gwen researches providers, finds emails, negotiates rates, redlines terms, and signs agreements.
• Gwen increased throughput from about 100 contracts a week to about 100 a day.
• In roughly eight weeks, Gwen completed 3,500 contracts versus 2,300 for the whole team last year.
• Human-done contracting costs about $1,500 to $2,000 per contract, versus about $70 with Gwen.
• Curative intends to do 10x to 20x more contracts using agents.
• Turner says back-office jobs are the most exposed to automation.
• He says Curative will keep investing in technical skill and relationship-driven roles.
• Even small employer contracts can be around $1 million.
• Curative allows Gwen to redline and sign standard contracts within guardrails without outside law-firm review.
• Turner says current models can already do all of Curative’s back-office tasks.
• He says enterprise AI spend could reach 2x to 5x salary-equivalent within three years.
• A major challenge in agent deployment is supervising exceptions as work volume rises.
• Turner says people underestimate model capability if they only tried older versions of ChatGPT.
• He says agents will eventually act on behalf of users, but trust will take time.
• Turner is bullish that AI will lift earnings in regulated industries like health insurance.
• Curative has not sold secondaries and says investors got 10x back via a dividend at the end of COVID.
• The last Curative round was $150 million at a $1.3 billion valuation.
• Turner co-founded Subcritical, a nuclear fission company, with his wife.
• Subcritical operates at 0.97 criticality and uses an external particle accelerator to drive the reaction.
• He says nuclear safety is mainly a regulatory problem, not an engineering one.
• A 300 MW plant would cost about $1 billion to build.
• AI could cut nuclear design teams from around 100 mechanical engineers to about 20.
• Curative’s design migration from Looker to Snowflake was reduced from a year-long project to a couple of months with an agentic workflow.
• Turner expects to keep the human care navigator role, but let agents amplify its output.
• A future role will be an agent supervisor to handle exceptions and approvals.
• Turner says current models can already do every back-office task at Curative.
• He believes trust in agents acting on behalf of users will take longer than internal workflow adoption.
• Turner thinks AI will massively boost earnings in regulated industries like health insurance.
• Curative has not sold secondaries; investors got 10x back via a dividend and still own shares.
• The company’s last round was $150 million at a $1.3 billion valuation, led by insiders.
• Turner co-founded Subcritical, a nuclear fission company, with his wife.
• Subcritical uses the energy amplifier concept: 0.97 criticality plus a particle accelerator.
• He argues nuclear’s main bottleneck is regulation, not science or engineering.
• A 300 MW deployment would cost about $1 billion and use project finance.
• AI could reduce nuclear design staffing from around 100 mechanical engineers to about 20.
• Curative’s Looker-to-Snowflake migration was largely handled by an agentic workflow.
• Curative’s care navigators remain human, but agents make them far more productive.
• Turner frames the biggest bottleneck as supervising agent exceptions at scale.
• Turner says people underweight model capability if they only tried older ChatGPT versions.
• He says agents will ultimately take on user-facing actions, but trust will lag behind internal adoption.
• Curative sees AI as a way to raise earnings in regulated sectors without worsening the product.
• Curative says it has not sold secondaries; investors got a dividend and still hold shares.
• The last Curative round was $150 million at a $1.3 billion valuation, led by insiders.
• Turner describes Subcritical’s reactor as a subcritical system driven by an accelerator and designed never to run away.
• A 300 MW Subcritical plant would cost about $1 billion and use equity plus debt.
• Turner says AI can compress nuclear design teams substantially, including CAD generation via code.
• Turner says Curative has become a smaller, more AI-amplified company in the short term.
• He says the next decade will reward people who combine technical skill and relationship-building.