Summary

Singh built TPG Axon partly because his daughter’s SMA turned investment performance into a race for research capital. Diagnosed at 18 months in 2001 and expected to survive only into her mid-teens, she made a two- or five-year delay unacceptable: “My day job was really there to pay for my night job.” He estimates the family ultimately committed about $150 million.
SMA offered an unusually asymmetric drug-development setup: a defective backup gene to repair, usable mouse and cell models, and a very wide therapeutic window. The response was portfolio logic—fund every credible path simultaneously because “we need to do things in parallel processing, not serial processing”—while paying for shared tests and letting contributors keep their IP.
All three “shots on goal” became approved drugs: an Ionis/Biogen antisense program, AveXis gene therapy bought by Novartis for $8 billion, and a PTC/Roche small molecule his daughter now takes daily. Singh calls that “incredible luck,” but it followed systematic de-risking. An early pitch used $5,000–$15,000 per patient-year to show that a rare-disease drug could still generate hundreds of millions in profit; he later estimated the SMA drug market at high-single-digit billions, with SMA drugs ranking among the top three for Roche, Novartis, and Biogen.
Speed came from engineering the entire path, not merely discovering a molecule. Singh’s group rallied 50 Nobel Prize-winning scientists to sign a letter, created political momentum, and funded a natural-history study so the FDA would not require dying children to enter placebo arms; he says the first SMA drug was perhaps the fastest FDA approval ever. “Kids are dying. We can’t afford to run a placebo trial.”
The result differs radically by timing: his daughter became stable after starting treatment at 12, while a child diagnosed and treated post-diagnosis from birth “will never show symptoms.” Stability does not restore lost muscle or nerve, so stage two is regeneration: one candidate is already in the clinic, and Singh expects three or four programs that might benefit SMA and broader muscle-loss populations over the next handful of years.
Rare disease is now commercially legible, but its capital stack still fails both before company formation and after approval. Singh wants neutral translational teams to move bespoke ideas “to midfield,” because grants and IP owners do not reward ruling out failures or collaboration; at the other end, $1 million–$2 million annual prices and cross-border rules can leave a child untreated even after a family raises $400,000. “The notion that we develop these drugs and there are kids that are dying ’cause they can’t afford them is pretty insane.”

Deep dive

1. A terminal clock turned investment success into research capital

Singh’s clock started in late 2001: a month after returning from Hong Kong and becoming global head of his Goldman group, 9/11 happened; a month later, his 18-month-old daughter was diagnosed as he watched the World Trade Center ruins from his office. A neurologist’s roughly 60-second explanation ended with a prognosis of progressive weakness and likely death in her mid-teens.
The scientific opening was narrow but real. SMA patients lack a gene producing a protein Singh likens to “fertilizer for muscles and nerves,” yet everyone carries a defective backup gene; copy count helps determine whether death comes within weeks or after 10–15 years. Repairing or amplifying that backup looked easier than replacing something wholly absent.
Development also had favorable experimental characteristics: researchers could make SMA mice and cell models, and natural variation suggested the protein had an unusually wide therapeutic window. Singh’s water analogy made the issue concrete—nearly anything becomes toxic at sufficient dosage—but with this protein “it’s pretty clear that you almost can’t have too much.”
Hope made delay morally intolerable: “The second worst thing … is to see their child suffer and die. The single worst thing … is … to find out that you could have done something about it.” Leaving Goldman to launch TPG Axon made every dollar earned immediately useful; Singh estimates that about $150 million of their own money was ultimately spent, while his father joked that he had developed “the world’s most expensive drug habit.”

2. Parallel processing produced three approved shots on goal

Rare disease then had almost no commercial constituency, and an innovative NIH translational-grant program selected SMA but stalled after leadership changes—“when you have acting directors, that means there is no action.” Singh’s diagnosis of “latency bias”: unfamiliar diseases attract fewer researchers, applications, and qualified reviewers, stretching the cycle beyond his daughter’s likely life.
Rather than fund academia serially, the family built a virtual drug company: hire scientists, assemble roughly 18 elite development advisers for working weekends, identify every missing experiment, then pay a lab or company to execute it. “If there are five interesting ideas, let’s do all five at the same time.” Academia mattered but would never develop the drug; “maybe it will work, maybe it won’t.”
They created multiple mouse and cell models and offered companies free testing while allowing them to retain all IP. It still took years to find a taker; Novartis eventually identified an abandoned 1970s anorexia candidate that looked “off the charts good,” although broader screening showed existing FDA-approved drugs would not be effective enough without toxicity. The platform also learned quickly what not to advance.
The portfolio covered three modalities: a Cold Spring Harbor scientist’s antisense program went through Ionis and was partnered with Biogen; PTC found the small molecule Roche advanced; and AveXis built gene therapy before Novartis bought it for $8 billion. All three won approval—“incredible luck.” An executive had laughed at Singh’s $5,000–$15,000 annual-price model; Biogen charged, in Singh’s recollection, about $730,000 a year for the first drug, and he estimates today’s market in the high-single-digit billions.

3. Evidence and coalition-building made the regulatory clock move

To create institutional momentum, Singh cultivated the NIH, FDA, senators, and congressmen; obtained a letter signed by 50 top scientists who were Nobel Prize winners; and ran full-page color ads in Washington political papers. The pitch was not pity—a parent “with tears on their face” gets little traction—but that SMA had a genuine chance of success, giving officials a fixable problem.
The pivotal regulatory design was a funded natural-history study: years of recorded care established the disease’s natural progression while the drugs were being built. FDA agreed those records could serve as a comparator, avoiding a blinded placebo trial in which half the children might deteriorate or die. It “saved a lot of lives and a lot of money” and prepared trials to move “ninety miles an hour.”
With persistence and help from his network, Singh says the first SMA drug was then—and might remain—the fastest FDA approval in history. His daughter began the first therapy at 12 and has been stable since; she now uses Roche’s daily treatment. When Patrick asked about a newborn treated post-diagnosis, Singh’s answer was categorical: the child “will never show symptoms.”

4. Plugging the leak made muscle regeneration testable

Drug speed could not reverse accumulated damage, so the family fought to preserve baseline: daily physical therapy, diet, and a warm therapy pool built in their apartment building. Before treatment, she still endured spinal fusion, hip problems, annual major surgeries around ages 10–11, and hospitalizations roughly three times a year for severe pneumonia. “Once it breaks, you can’t just put it back together again.”
Singh’s boat metaphor defines the second-stage thesis: bailing is futile while the hull still leaks, but after repairing the hole, removing accumulated water can work. SMA-specific drugs address the underlying loss for roughly 20,000 children; his programs now aim to regenerate muscle—easier than nerve—and strengthen surviving nerves, potentially serving muscular dystrophy and other weakness beyond SMA.
Muscle was surprisingly underdeveloped—Singh reckons perhaps six people in the country truly understand it—because trial populations were either people with unsolved degenerative disease or elderly, weakening patients whose noisy decline obscures moderate benefit. Treated SMA changes the experiment: the leak is fixed, so incremental effect becomes measurable. One program is already in clinic, though “not the best one,” with three or four more expected to come forward.

5. Community converted sorrow and rage into sustained action

The outcome is stability, not invulnerability. His daughter nevertheless became “the rock star of the family”: straight A’s and summa cum laude at Yale, three of the 11 awards Singh cited, class speaker, and now a Cornell PhD student and married. Singh keeps pursuing regeneration because she remains fragile, and aging can turn fragility into new risk.
Her trajectory sharpened his indictment of institutions that misread disability. Despite accessible buildings and Singh’s board relationships, major New York private schools refused her; one said she was not bright enough, while another requested medical records and a doctor conversation that Singh called illegal. All Souls initially suggested a cerebral-palsy school—wrong disease—then admitted her, and she thrived.
In a nursery-school sermon, Singh named three adaptations: borrow strength from community; stop grieving only the imagined life and recognize that different sources of happiness need not reduce their magnitude; and channel anger and sorrow into useful work. “In life there are no do-overs, there are no retractions, there are no appeals. You either find joy in the life you have or live a life without joy.”
Finance and science ultimately formed a family circle: his mother had worked on the faculty at Columbia, the leading center in the work; his father had worked at Roche; and a Roche scholarship paid for Singh’s college. Yet his answer about kindness was smaller than the institutions or capital: after the diagnosis, hugs supplied borrowed “strength and energy” to stand up and move again.

6. Rare-disease capital needs both a bridge and a price constraint

Singh now receives calls every few weeks from parents who want impact rather than donations “like throwing water on sand.” At Jackson Laboratory—whose mouse-model work his SMA effort helped make freely available—he is building a rare-disease institute, the “Ghostbusters of rare disease”: families call, scientists assess what can and cannot be done, and a brain trust supplies a game plan without requiring the work to remain inside the institute.
Commercial appetite improved once rare-disease profits became visible, but bespoke projects still need de-risking before biotech capital engages: “You don’t have to get it to the ten-yard line anymore, but you still have to get it to midfield.” Companies optimize their own IP; academics optimize grants, which do not reward parallel screens or proving an idea fails. Modest translational capital could close that early gap.
The coordinating institution must also be “Switzerland.” Academics resist collaboration, companies guard IP, and universities increasingly seek royalties; a neutral NIH-like or nonprofit team can select a few diseases, align experiments, and serve as honest broker. Singh’s point is not that government failed wholesale—the original NIH idea was “pretty cool”—but that institutions can organize small, practical discovery programs more intelligently.
Success created the opposite failure at access. Singh calls $1 million–$2 million annual prices “bonkers”; at an $800,000 first-drug cost, a major center such as Columbia needed tens of millions in working capital merely to stock doses. In Uzbekistan, one family raised $400,000 through community collections yet found neither someone to take the money nor a practical route to treatment. He wishes his funding had imposed price limits and suggests reserving windfall profits for charity care.

After Landing a16z Backing, They Want to Rebuild Advertising with AI | A Conversation with 00s-born CEO Jason Hu Terence Tao: Hardest Problems in Mathematics, Physics & the Future of AI | Lex Fridman Podcast #472