Summary

Bertolini’s core call at Oscar Health: eliminate employer-sponsored insurance and move America to individual purchase on ACA rails — the defined-benefit-to-defined-contribution pension transition replayed in health benefits. Oscar covers ~2 million ACA lives today; he sizes the endgame at 125 million people: 79M small/mid-group employees after five straight years of double-digit rate increases, 20–30M gig workers via “hour banking,” and another 80–100M as large employers tip in. The pain point is real: “two out of five Americans borrowed $74.9 billion” last year just to cover out-of-pocket healthcare costs.
Oscar’s claimed cost edge: the first new insurance platform built since 1972 (the industry largely uses old HealthEdge), cloud-native with one version of truth in data, and “almost two dozen” LLMs already cutting operating costs — enabling counterintuitive products like actively recruiting diabetics: uncontrolled costs $1,600/month vs. $900 controlled, so “you can invest $699 and get a return.”
Key regulatory catalyst-slash-risk: the ICHRA-style CHOICE provision was in the House bill but was stripped by the Senate over a ~$30B tax score; further resistance comes from corporate benefits staffs, fee-collecting consultants, and 51 contradictory regulators.
Why healthcare stays broken structurally: 760,000 physicians run on cash accounting, ~13,000 hospitals and facilities on revenue, 1,300 insurers on margin — “it’s a wonder people get care at all” — plus Machiavelli’s warning about initiating a new order of things: “that’s where I’ve lived my whole career. I am an iconoclast.”
Aetna proof points for the model: relaxing hospice rules (no forced surrender of curative hope) moved uptake from 18% to 75% and cut end-of-life costs 75%; raising frontline minimum wage to $16 and eliminating out-of-pocket healthcare costs (a $75M program announced at JPM 2015 without board approval) preceded the stock going $39 to $80, part of a 652% eight-year TSR — two-thirds of it multiple expansion.
The operating philosophy: “while I thought I was right 100% of the time, I’m really only right about 35% of the time” — form a hypothesis, build a management process, vector like sailing; hire for curiosity and courage; “we are here to help ordinary people do extraordinary things.”
The human story framing it all: his son’s 34-cases-ever lymphoma was treated with deliberate graft-versus-host disease and FDA compassionate-use [likely Omegaven]; Bertolini’s chapel bargain — “me for him” — followed by a broken neck exactly one year later, then 18 years of level 7–10 pain on one hour of sleep a night, with the pain ending in six months after a doctor evaluating him for assisted suicide at Dignitas diagnosed neuroplastic, not intractable, pain.
The Bridgewater cameo: chaired Dalio’s exit (deal closed 4:15 a.m., July 15, toasted with Pappy Van Winkle), rode shotgun for Nir Bar Dea [likely] rather than take the CEO seat, and found the culture “so institutionalized” that changing the management process was the real struggle — “the investors who run the business and make all the money, they call the shots.”

Deep dive

1. Memento mori: a broken neck, a frozen river, and a ring

The accident, as told: skiing into the woods, caught an edge, hit a tree — five broken vertebrae, macerated brachial plexus, nerve root disconnected from the spinal cord — then a 60-foot headfirst slide into a river, two hours to extract, last rites on the medevac, five days in a coma. The nurse giving him a sponge bath “went running from the room and said, ‘Oh my god, he’s alive.’”
The crazy detail: the ice and cold water saved him. In cases like his, spinal-cord swelling can create damage; he said the ice and cold water prevented his spinal cord from rupturing.
A friend sent him the memento mori ring he still wears: “Remember you will die… every day when you wear this ring, look at it and ask yourself, am I spending my time with the right people doing right things? And if you’re not, run.” The Dalai Lama’s version, when Bertolini asked how to handle frightening people: “in your heart you need to find compassion for the journey they’re on… but physically run. And he laughed.”

2. Detroit sandlot to Ford axle line to Cornell — merit, bullies, and a photographic memory

Oldest of six kids in seven years, 1,000-square-foot house, one bathroom, sleeping in a breezeway; sandlot baseball where captains picked teams and “if you didn’t get picked, that meant you weren’t good. There were no participation trophies.” The neighborhood lesson that stuck: “there is no good bully and the only way to beat a bully is to challenge them.”
After flunking out of college he worked rear axle differential housings on the Mercury Bobcat at Ford, fought his union steward over work rules, and asked “How do I get your job?” — the answer (“you need a degree, kid”) sent him to an accounting degree, then a GMAT taken on a lark, then Cornell’s MBA (chosen partly because “you can drink beer and smoke pot down in the gorges”).
On the photographic memory: “scary. It makes you lazy.” In business it made him “a forensic scientist looking at spreadsheets to see which cells were changed in order to make the numbers work… it actually glows on the page for me.”
The career pattern after his first job: one job search ever; everything since has been inbound calls to “fix broken things and build new things, not to make the trains run on time” — first HMO turnaround, New York Life, then Aetna, which was “losing a million dollars a day” when he arrived.

3. The fixer playbook: Aristotelians, 35% hit rates, and watermelons on the floor

His first company nearly died in three years, teaching the distinction he still uses: Platonists vs. Aristotelians. “Platonists are people who think the great idea, communicate it and consider their job done… Aristotelians actually have to translate it into real work” — and being an Aristotelian is “far more powerful.”
The epistemic core: “while I thought I was right 100% of the time, I’m really only right about 35% of the time. But it’s just the issue of making a decision and moving forward” with a hypothesis and a management process. His metaphor: “it’s a lot like sailing. You don’t sail from one point to the next. You actually have to vector.”
At Oscar his first ten weeks were nothing but deep dives — when Josh Kushner called asking when things would happen: “I have screwed up everything this company has screwed up at least twice myself… I can’t fire everybody. I want to bring them to a place where they can succeed.” What he expected to be low-hanging fruit “were actually watermelons rolling around on the floor.”
How ordinary people do extraordinary things: “all they need to do is believe they can do it” — via story. His at Oscar: “we are a pirate ship filled with cannons among Spanish galleons filled with gold. All we have to do is take their business… They won’t go where we will go. And that’s how we win.”

4. The 80-year accident of employer insurance — and the case for killing it

His cliffnotes history: health insurance began ~80 years ago as a wage-control workaround (tax-free benefits instead of raises), then the Hill-Burton Act (he dates it 1952) built hospitals in every American community. “If you create enormous capacity and you give people no sensitivity to price… you have created a great inflationary monster.”
The prescription: “we need to eliminate employer sponsored insurance” and let individuals buy on the ACA marketplace — explicitly the DB-to-DC pension transition, with brokers playing the financial-advisor role. Large employer networks are “inversely cost effective”; in the ACA, narrow networks by market let people pick “my network, my benefit plan for me and my family,” bronze to gold.
The small-group death spiral: one pregnancy gives a small employer a double-digit increase, everyone else pays 15% of a rising sum — “five years running, small group employers have been paying double-digit rate increases” and are now dropping coverage. Meanwhile “two out of five Americans borrowed $74.9 billion” from banks for out-of-pocket costs last year.
Patrick’s steelman — smart skeptics say sick people get hurt in a consumer market. Bertolini: “It’s exactly the opposite. Consumers when educated can make a good choice because it’s about them” — and portability means “as their circumstances change… they can keep that network and that plan.”

5. Oscar today: a 1972-vintage industry meets a cloud-native stack

State of the union: ~2 million lives, all ACA, expanding markets, a condition-explainer chatbot about to launch, and “almost two dozen large language models on the back end” that have already reduced operating costs — on “the first new platform for health insurance built since 1972” (he’s personally modified the incumbent HealthEdge platform three times in his career). One version of truth in data is what makes AI at scale possible.
The diabetes product as the model in miniature: Oscar encourages diabetics to join — “counterintuitive in the insurance industry, asking sick people who cost $1,600 to $1,700 a month to join your plan.” An uncontrolled diabetic runs ~$1,600/month, controlled ~$900: “$700 opportunity… you can invest $699 and get a return,” via wearable glucose sensors and self-paced digital tools.
The plumbing that makes it work: claims with diagnosis codes go to the government several times a year; plans carrying more risk get paid by plans carrying less through risk adjustment, so attracting sick-but-engaged members is economically rational.

6. From 2 million to 125 million lives — and the premortem

The TAM math: 79 million in small/middle-market groups who’d be better served by an ICHRA-style defined contribution; 20–30 million gig workers who could buy in via “hour banking” (accrue hourly set-asides, aggregate monthly, buy a policy); 80–100 million more as large employers move part-timers and then everyone else. “It’s a massive market” — he says 125 million total is servable.
The consumer pitch: take the employer’s defined contribution, buy a plan built around your actual year — “Are you going to have a kid this year?… You have money left over most often” for out-of-pocket costs, dental, vision, life, disability. “All of a sudden, it’s my plan. It’s my choice… which Americans have in every other part of our economy.” Guardrails stay: the money must buy healthcare, not “beer.”
Patrick’s premortem — why might it fail? Bertolini’s list: large employers’ hundreds-strong benefits staffs become redundant; consultants lose annual fees; and regulation — the CHOICE provision was in the House bill but the Senate pulled it over the tax scoring of HSAs, “it saved 30 billion so they pulled it out” (he argues the score is wrong since the benefits are tax-advantaged today anyway). Indiana has already made the changes at the state level; brokers, he thinks, can be converted into allies.

7. Why the system fights back: three accounting models and Machiavelli

Size cuts both ways: incumbents took care of 54 million (his Aetna scale) but run ACA as a C-team subsidiary — some are exiting because they lazily used commercial networks instead of narrow ones — while Oscar has “the A team on this product for these people every day.”
The regulatory maze: 51 regulators (50 states plus federal) actively contradicting each other — e.g., national carriers fined per case when one state forbids abortion coverage another state’s employer offers. His fix: “it would be far better if we just said to the states, you run your own exchange.”
The structural diagnosis worth remembering: 760,000 physicians on a cash basis, ~6,900 hospitals plus ~6,900 facilities on a revenue basis, 1,300 insurers (and pharma, biotech, DME) on a margin basis — “you put those economic models together, it’s a wonder people get care at all.”
The quote he’s kept under his blotter his whole career, Machiavelli to the Prince: “there’s nothing more dangerous… than to initiate a new order of things,” because incumbents have the laws and experience, and the beneficiaries won’t believe until they experience it. “And that’s where I’ve lived… I am an iconoclast. Nothing should remain the same unless it can stand on its own.”

8. AI as grenade — and the hospice rule change that paid for itself

On clinician knowledge decay: the best-prepared doctors are fresh residents, and “if they were to read two journal articles every night for a year, by the time the year passed, they would be 5 years behind.” AI’s role is better-informed consumers rather than deference: “the mental model that exists inside your head… is the most critical tool you have.” His two executive filters: curiosity and courage.
The Aetna hospice experiment, born of his son’s ordeal: end of life is where “60% of your lifetime costs are spent,” and rules forced patients to renounce curative care to enter hospice. When Aetna dropped that requirement — over internal warnings that “you’re going to break the company’s back” — hospice uptake went from 18% to 75%, and costs were 75% lower in end of life, alongside letters about how much better final days were. “Changing the rules about how we do these things — AI can help us in these places.”

9. Eric’s story: engineered graft-versus-host, an Austrian drug, and “me for him”

His son’s diagnosis at 16: T-cell gamma delta lymphoma — only 34 people, all men, ever diagnosed, six months to live. Bertolini quit his job; Dr. Eva Guinan [likely] at Boston Children’s proposed deliberately inducing graft-versus-host disease so the transplant would attack the cancerous T-cells. Brother Peter was “the perfect mismatch.” The cancer died; GVHD nearly killed Eric — skin sloughing, 140 lbs to 80 lbs, bleeding his total blood volume daily, Bertolini donating his own blood, living in the room, doing hour-long rounds with his Harrison’s textbook and PubMed.
After conceding to hospice (“Eric, game over”), Bertolini refused to accept starvation as unsolvable: a resident’s call to a Hopkins nutrition professor surfaced [likely Omegaven], a non-soy omega-3 lipid from [likely Fresenius], obtained in Austria and not FDA-approved — obtained via a single-patient compassionate-use exception, hand-carried from Austria by the Fresenius CEO. Four weeks later Eric’s hair grew back; he came home February 18, 2003. Two years later he got Bertolini’s left kidney, laparoscopically.
The bargain at the chapel: scapular and rosary on the altar — “Me for him. Let me have his disease. Let me have his pain… I’ve lived a great life.” Eric came home February 18, 2003. “February 18, 2004, I hit a tree, break my neck.” His answer to Eric’s “why are we both here?”: the Neil deGrasse Tyson lottery framing, plus “I’m just spare parts. And by the way, you can’t have the other kidney until I’m gone.”

10. Eighteen years as a business terminator — then a six-month cure

The pain years: “level 7 to 10 pain, sleeping one hour a night for 18 years.” His account of the neurology: “the right side of my brain literally shut down. So my level of empathy was pretty low. Left side of my brain was a machine.” Patrick’s summary he accepts: “you were just like a business terminator.” The machine at work: daily 6:30 a.m. reserve runs at Aetna, where every 50 basis points of trend was $980 million of margin — a problem meant everyone in his office at 8:00.
The turn: he had scheduled assisted suicide at Dignitas for February 15, 2023 (he wanted the 18th — the accident’s anniversary — but it fell on a Sunday). The doctor evaluating him for intractable pain instead said: “You don’t have intractable pain. You have neuroplastic pain and we can fix it.” His 42 pain centers had wired together from chronic pain — “you have a baseball hitter’s brain on pain,” zero latency, any touch registering as pain. Treatment: CBT into childhood trauma, transcranial magnetic stimulation to disconnect the pain centers, VR repatterning. Six months. Out of pain — after a prior failure with seven days of 50mg/hour IV ketamine.
The aftermath was not relief but rage — “my brain betrayed me. I thought I was smart” — then “incredible remorse for my lack of empathy… I was patient, but I was not tolerant. And it should be the other way around.” Then emotions returned wholesale: “I’m watching dog commercials. I’m crying.”

11. Yoga journals to a $16 wage: the Aetna experiment that moved the stock $39 to $80

Cranial sacral therapy helped him come off narcotics; yoga (“yoga’s for girls,” he protested) led him into the Gita and the yamas/niyamas. When his CMO called workplace yoga “voodoo medicine,” they ran a bicoastal double-blind study, 795 employees: the highest stress quintile spent $2,500 more per year on healthcare; after 12 weeks, cortisol and heart-rate variability dropped 50%, and actual healthcare dollars — not trend — fell 7.5% the next year.
Then the journals landed: “Okay, Mr. Big Shot CEO… the problem with their stress is you.” Frontline staff at ~$12/hour, 81% women, families on food stamps and kids on Medicaid, calling bill collectors on break. His response: minimum wage to $16, and for the 7,100 employees under 300% of federal poverty, elimination of all out-of-pocket healthcare costs — a $75M program announced at the January 2015 JPMorgan conference in front of holders of 240 of 340 million shares, without asking the board: “ask for forgiveness and not for permission.”
The payoff, as he tells it: spend grew to $125–130M/year and “our stock price goes from $39 a share up to 80 because we then gave permission for our employees to care for our members like we were caring for each other.” His last five years at Aetna “were joyful because I didn’t have to tell people to get to Sandy Hook… it all just happened.”
The CFO-proofing method: put the soft benefits (presenteeism, empathy) on a page and discount them 90% — they still equaled the hard savings — then have the leadership team rank the top five risks and decide as believers, “not work by what the spreadsheet tells us.” His verdict: “the spreadsheet is the worst invention ever created for business… we actually create a belief system that’s not real and then we manage to it.”

12. Never count anyone else’s money: the Jerry lesson, “no margin, no mission,” and a 652% TSR

The formative story: at 14, discovering coworker Jerry made $4.25/hour to his $1.25, he demanded a raise; his father fired him, then explained at dinner that Jerry had a family and limited prospects — “I have an obligation to Jerry” — and rehired him at $1.00/hour. The lessons, verbatim: “Never count anyone else’s money… when you have it, make sure you share it… Never threaten to quit a job unless you got another one to go to.”
His compensation doctrine for everyone who works with him: salary pays the bills, bonus pays for the balls, “and your stock is never to be touched. It’s part of generational wealth.” His ~3 million Aetna shares went largely into the Anahata foundation (heart chakra), which gives away $12–15M/year — “a great way for the federal government not to get it.”
Sister Joyce of the Daughters of Charity settled the for-profit question for him: “No margin, no mission… the only difference between a for-profit and a not-for-profit that are well-run is the for-profit pays taxes.” At an investor conference a young woman stood up wearing it on the back of her T-shirt.
The market-structure observation investors should note: of Aetna’s 652% total shareholder return over eight years, “only a third of it related to earnings and two-thirds… to the price earnings multiple,” i.e., the street’s belief in the franchise — “and that was driven by our employees. We created 2,000 new millionaires when we sold the company at CBS.”

13. The Bridgewater interlude: structuring Dalio’s exit, then declining the throne

Fall 2018: Ray Dalio, mid-70s, tells him “I really worry that the company will go to hell in a handbasket when I’m gone… I think they’ll just spend everything if I don’t” — “the typical entrepreneur founder kind of thing.” Bertolini joined the board that February, built committee structures and governance, then became chairman at Dalio’s request.
The deal closed at 4:15 a.m. on July 15 after marathon sessions (“no detail too small”), toasted over Zoom with a bottle of Pappy Van Winkle. When Dave McCormick left to run for Senate in Pennsylvania (“for reasons I’ll never understand”), Dalio floated Bertolini as CEO; he declined the full job — “I don’t know the business like other people do” — and instead did an 18-month deal to “ride shotgun with Nir [likely Nir Bar Dea] until he’s ready.”
His candid read on the firm: “the culture is so institutionalized… causing any changes at the management process level about how we work was really difficult… the investors who run the business and make all the money, they call the shots, and that’s how it works in those places.” When it ended, Josh and Mario — with whom he’d been meeting an hour a week for years — asked him to run Oscar.

14. Why Oscar: 1991 acetates, the zip-code thesis, and the angora sweaters

Cleaning out boxes after Aetna, he found acetate slides from 1991: human genome project, virtual healthcare, the individual market. “Oh my god, we haven’t accomplished anything in 32 years… back then it was 14% or 15%” of GDP, now 20%. “This isn’t going to end well.” The CVS deal was his first attempt at the fix; Oscar — small, nimble, the pirate ship whose original cohort of insurtech peers “are all gone” — is the second.
The engagement insight, from IDEO work with George Blankenship (Tesla/Apple stores): patients don’t describe themselves as disease states. The question that works: “What is it about your health that gets in the way of the life you want to lead?” — a diabetic with foot neuropathy wants to walk grandchildren to the park, not resemble “the person on the front of Men’s Fitness magazine.” He notes 41% of Americans have functionally given up (“my mother died of sugar, I’m eventually going to get it”).
The social-determinants epiphany came from crash-coursing nine months for grand rounds with Sir Michael Marmot: “your zip code is far more important than your genetic code.” The case that proves it: a 72-year-old in Camden, NJ — $2.7M in healthcare costs in one year, ER visits “more than one a day” — whose house was full of blankets and sweaters made of angora, to which she was allergic, with the thermostat at 62. “Buy her new sweaters, buy her new blankets. The next year, one ER visit.”
On knowing what to do at all: Bill Taylor’s advice — cultivate the network, keep feeding “the mental heuristic in your brain” — plus the pandemic’s gift: it proved virtual care works, converting even Andrew Weil [likely], who’d insisted “we have to put our hands on people.”

15. Courage, humility, and the kindest thing anyone did

His definition of courage: “moving forward without perfect information. 99 out of 100 people sitting around a table are looking for somebody else to make a decision.” The parable: his own electrolyte crash after food poisoning, worsened by Smartwater, solved by a UK-trained family doctor with a $1 Diet Coke and a hypothesis — “we don’t have all these machines that go beep… it’s based on having a good hypothesis and knowing how to test it. That’s how I work.”
The humility correction he credits to Ron Williams, 2003: stop leading the witness with pre-scripted questions — “why don’t you actually listen to the answers… you’ll be amazed at how much smarter you are. It changed my style completely.” His advice to his late-1990s self: “a lot more humility… when you tell people what to do and you stumble, the knives come out.”
On unfinished business, an honest non-answer: despite Harlem Grown (one farm, now 25 across Harlem elementary schools), rebuilding Harlem libraries, and reforesting the five boroughs — “I’m like unsatisfied and I don’t know why.” On endurance: “never waste a good crisis… every crisis you face, you can’t go back. You can only go forward.”
The kindest thing: the doctor’s McDonald’s-fries summonses whenever Eric’s case took a dark turn — “don’t let them convince you that he’s dead. Here’s what I would do next” — advocacy that cost the doctor the leadership of the transplant program as the hospital pushed DNRs and even tried to emancipate Eric as a minor. “He’s alive because of my engagement and my advocacy… and her partnership in it. That’s the kindest thing that anybody’s ever done for me.”

August 2025 Random Ramblings Conversation with Guangmi: Model Divergence, L4 and Mining Windows