The episode kicks off with a punchy cold open in which Fred Turner delivers the headline figures: 206,000 COVID tests in a single day, $5 billion in total revenue, a team growing from 7 to 7,000 in nine months, and a bold prediction that Anthropic could be worth $10 trillion. Harry then delivers enthusiastic sponsor reads for AlphaSense (AI-powered market intelligence), MongoDB (the AI-native database platform powering 75% of Fortune 100 critical apps and 40 million ElevenLabs agents), and Framer (an AI website builder offering 30% off Pro annual plans at framer.com/20vc). The cold open efficiently signals to listeners that this episode will span COVID-scale logistics, AI transformation, and audacious long-range forecasting.

Fred's first company, TL Biolabs, began after he won the UK National Science & Engineering Competition and was approached by a farmer to test his cows' genetics for muscle yield and milk production. He applied to Y Combinator six days before the deadline — on the advice of a YC alum he met at an AgTech conference in San Francisco — and joined the Summer 2016 batch, raising a $1.65 million seed from Andreessen Horowitz's bio fund. But a Series A proved impossible once investors ran the TAM: 100 million US cows at $15-20 per test caps the market at $1.5 billion. The team pivoted the core DNA-testing technology first to high-throughput STD testing, identifying antibiotic resistance in sexually transmitted infections as a compelling and growing problem. Fred notes STD rates are continuing to rise despite people having less sex — a detail that generates genuine surprise from Harry. A second pivot followed into sepsis diagnostics, a market Fred describes with striking precision: bacteria in the bloodstream trigger the immune system to destroy the body's own organs, and every untreated hour raises mortality by 12% ^{[1] — Fred Turner "Sepsis: 12% mortality increase per hour untreated: Every hour sepsis goes untreated, a patient's mortality risk increases by approximately …" 14:14 Play} . The company renamed Shield, raised a Series A, built working prototypes, and pursued FDA approval — before a strategic acquirer's CEO killed a signed term sheet after three weeks of document work. With three weeks of cash left, Fred sold the company's hard-won CLIA lab license for $150,000 and began winding down. Five months later, that same license would have been worth $27 million ^{[2] — Fred Turner "I sold that license to a company in San Diego for $150,000 to pay some of the creditors. And then 5 months later, acquired a company in Sou…" 15:45 Play} .

The COVID chapter opens with Fred watching infection rate data on Twitter in mid-February 2020 and realising the trajectory was worse than anyone in public was admitting. With a COVID test already developed in his CSO's spare time but no lab license to run it, Fred scrambled — eventually striking a 50/50 JV with a sports doping lab in San Dimas, a small city best known as the setting of Bill & Ted's Excellent Adventure. Their first government customer was the San Dimas sheriff's department, followed quickly by the City of LA — a contract that came about because longevity investor Laura Deming tweeted about Curative's testing capacity and the Deputy Mayor slid into her DMs. To fund scaling, Curative collected daily invoices from LA with net-1 payment terms, sending someone to City Hall each morning to collect a check. The largest single contract — worth hundreds of millions — was Florida's statewide nursing home testing programme, which every other lab said was impossible to deliver ^{[1] — Fred Turner "Hundreds of millions single contract: Curative's single largest contract — with the state of Florida for nursing home testing — was worth h…" 27:23 Play} . Fred explains the orthogonal supply chain philosophy: use no consumables your competitors are fighting over. Curative sourced swabs from electronics testing vendors, replaced magnetic bead extraction with filter plates that could be manufactured at scale, and built from scratch rather than optimising existing processes. Hiring was done with 5-minute interview slots in socially distanced parking lot queues. The peak came in December 2020: 206,000 tests in a single day ^{[2] — Fred Turner "206,000 tests in a single day: Curative's peak single-day COVID testing volume was 206,000 people, reached in December 2020, just 8 months …" 30:55 Play} , with 7,000 employees. The margin profile was brutal — highly profitable at surge peaks, deeply loss-making in the troughs between variants. Over three years, total revenue hit $5 billion ^{[3] — Fred Turner "$5 billion total COVID revenue: Curative generated approximately $5 billion in total revenue from COVID testing over a 3-year period." 31:46 Play} . After paying out a 10x dividend to investors, roughly $500 million was reinvested into the health insurance company.

After ruling out lab testing (TAM too small at ~$30B), hospitals (too fragmented a payer mix), and primary care chains (all roads lead back to the payer anyway), Fred concluded that the insurance company is the only entity in US healthcare with true system-wide leverage. If you control the dollars, you control behaviour. He began seriously exploring health insurance in late 2021, with the business formally launching after obtaining the necessary licences. Fred's structural critique of US healthcare is pointed: a handful of mega-payers and a small number of consolidated hospital systems have created a market where neither side has real alternatives. A hospital-affiliated primary care doctor gets paid on average double an independent equivalent — not because they're better, but because the hospital system uses its market power to bundle access. When there are only four dominant payers and heavily consolidated health systems, the usual mechanisms of competition break down entirely. Fred argues the only fix is breaking up the negotiating units — more smaller payers, more smaller health systems — to restore a functioning market. Despite the dysfunction, he notes the US still has the most advanced medical capabilities on earth, and if you are seriously ill, it remains the best place in the world to receive treatment.

Fred opens this chapter with the admission that if he had known AI was coming when he designed Curative's health insurance operations in 2022, he would have built the entire business differently. Health insurance, he explains, is fundamentally a bits-moving business: the only physical product is a small plastic card; everything else is data flowing through a marketplace. That makes it uniquely susceptible to AI transformation. The first department to go to zero people was credentialing — the labour-intensive process of verifying that every doctor joining the network holds a valid license and is not being sued for malpractice. A Claude-powered agent now visits medical board websites, verifies licenses, reads transcripts, and checks malpractice databases end-to-end, cutting turnaround from 2-3 months to 12 hours and cost from $50 to $0.20 per credential ^{[1] — Fred Turner "Credentialing: $50 to $0.20: AI agents cut Curative's doctor credentialing cost from $50 to $0.20 per credential and turnaround time from 2…" 43:02 Play} . Claims processing has also been substantially automated. The most elegant solution Fred describes is in underwriting intake: rather than forcing brokers to reformat files into a standard template, Curative now accepts any format and instructs an AI agent to write Python code that converts the file on the fly. The code is used once and discarded — throwaway single-use scripts that eliminate the need for hundreds of data-entry staff.

The SaaS-is-dead discussion crystallises around Salesforce. Fred matter-of-factly states that Curative cancelled its $600,000 annual contract because a vibe-coded internal CRM, built in approximately two months, now works better for their specific workflows and hosts their AI agents natively. Nobody was using Salesforce anymore. The broader thesis is structural: large SaaS platforms are built to serve all customers generically, which means they serve no customer perfectly. They require dedicated administrators — Curative had a full-time Salesforce admin — and are notoriously difficult to extend with custom features. An in-house vibe-coded system can be tailored precisely, updated instantly, and integrates with every internal agent without additional configuration. Fred acknowledges maintenance is a genuine challenge and that this approach works best when you have strong technical resources. He estimates Curative will cut approximately 80% of all SaaS spend this year, with an internal tracking slide devoted to mapping every contract renewal date to the person responsible for cancelling it ^{[1] — Fred Turner "80% SaaS spend cut: Curative is cutting approximately 80% of its total SaaS spend in the current year, replacing legacy software with inter…" 47:13 Play} . He draws a distinction between infrastructure-layer software (Sentry, Slack) which has deeper switching costs, and application-layer software like CRMs and claims systems, which is now squarely in the crosshairs.

Harry asks whether legacy health insurers are structurally disadvantaged by their inability to rapidly build AI agent capabilities. Fred's answer is blunt: yes, the biggest insurers will struggle. It's not primarily a technology problem — they could hire the engineers. It's an organisational problem. If you have 100,000 employees and AI can replace 50,000 of them, the executives whose fiefdoms just got cut in half will resist at every turn. Change will happen, but over ten years rather than three. Curative, building from a smaller base with a technical culture, can compress that timeline dramatically. The conversation turns to the structural economics of health insurance: 85% of every premium dollar must by law go to care costs, meaning the only way to grow profits is to grow total spending — a perverse incentive Fred traces directly to the ACA's medical loss ratio provision ^{[1] — Fred Turner "Insurance: 85% MLR mandated by law: US law mandates that health insurers spend at least 85% of collected premiums on care, capping their pr…" 49:13 Play} . AI changes this equation: it can deliver better margins within the 15% admin budget rather than requiring increased total spending. Fred then makes his boldest prediction: Anthropic could be worth $10 trillion. His basis is direct — Curative's Anthropic spend has grown 6x every month for the past 6-7 months, from tens of thousands to millions of dollars, as the team keeps discovering new use cases faster than they can deploy them. Gwen, their provider contracting agent, sends 15,000 customised emails every day and follows up relentlessly — the thing Fred says humans can't scale because most salespeople give up after three attempts.

Harry asks whether current layoffs are genuine AI-driven restructuring or simply corrections for over-hiring in 2021-22. Fred says it's both, but is clear that structural change is real. At Curative, back office workflows are being automated one by one, and he expects the company to shrink from 650 to around 400 employees in the short term before growing again as the business scales. He draws a sharp distinction between the two categories of work that he believes will remain human: deep technical skills (engineers who architect and oversee agent deployments) and authentic relationships (the broker golf dinners, the hospital-system conversations that require physical presence and trust). Clinical roles — care navigators who stay with members throughout their healthcare journey — will grow linearly with membership but become dramatically more productive as agents handle follow-up tasks at population scale. Fred introduces the concept of the Agent Supervisor as the emerging critical role: as agents do 10x the work, they also generate 10x the exception requests requiring human judgment, and managing that queue effectively is a skill that doesn't yet have a name or a training pipeline. He ends with a provocative vision: current-generation models can already do every back-office task at Curative — the only bottleneck is deployment.

Fred didn't set out to start a nuclear company — he went looking to invest in one, figuring his expertise in navigating highly regulated industries could accelerate an existing player. What he found instead was disappointment: almost everyone in nuclear was treating it as an engineering challenge, when the real barrier is regulatory. Safe nuclear reactors have existed since the 1960s; the engineering is solved. The problem is that the anti-nuclear push of the 1980s created a regulatory environment requiring absolute guarantees under once-in-a-million-year scenarios, which traditional criticality-balanced reactors cannot provide. Fred then stumbled upon the energy amplifier concept developed by Carlo Rubbia, the Nobel laureate and former CERN director, in the late 1980s. The design operates permanently at 0.97 criticality — always below the threshold at which a chain reaction sustains itself. A particle accelerator provides the extra neutrons needed to generate power. When the accelerator turns off, the reaction stops immediately. No matter what you do to the reactor, even 10x the accelerator power, criticality cannot be reached. This removes the fundamental guarantee problem from nuclear regulation. The only active construction of such a system is in China, based on a US design abandoned after Fukushima. Fred and his wife co-founded Subcritical to fill that gap, estimating each 300-megawatt deployment at approximately $1 billion in construction cost, financed through infrastructure-style equity-and-debt structures. He predicts coal will largely disappear from the energy mix over the next decade, and sees AI-assisted mechanical design as a key enabler for accelerating Subcritical's development.

Harry's quickfire round opens with the question of what Fred has changed his mind on most in the last twelve months. His answer is striking: a year ago he would have said there are workflows current models simply cannot handle. Today he believes every single back-office flow at Curative is within reach of today's models — it's purely a matter of deployment, configuration, and policy-setting. On Europe's competitiveness problem, Fred proposes something almost politically toxic on the continent: regulations should come with a tax penalty. Every new regulation passed should impose an additional burden on the country, creating a real cost for adding rules without reforming existing ones. The incentive to regulate addictively needs a counterweight. Fred's final reflection on the best advice he has received is a lesson from COVID-era Curative: the more radically different perspectives you can gather on a problem — former military logistics experts, Silicon Valley developers, research scientists — the closer to ground truth you can get. No single perspective is the truth; the truth lives in the synthesis. Harry signs off with genuine enthusiasm, calling it the most wide-ranging conversation he has ever done on 20VC, followed by a second full sponsor read for AlphaSense, MongoDB, and Framer to close the episode.