Summary

Greenoaks is organized around one hunt: the 10–15 founders each year who might build future S&P 500 companies, then giving them the firm’s full attention and capital. Mehta wants “the vital few” rather than broad deal coverage; O’Shaughnessy’s introduction credits this concentration with more than $13 billion of profits and a 33% net IRR across nearly $15 billion of AUM.
The product signal Mehta prizes is a “jaw-dropping customer experience” created by breaking technical or operational trade-offs competitors avoid. Coupang spent years making 12-to-24-hour delivery reliable across warehouses, routing, packaging, inventory and last-mile camps; market retention was in the 30s while Rocket cohorts were in the 60s, and mothers said, “If you took this away from me, I don’t know what I would do.”
For Mehta, exceptional founders rarely stay trapped in bad businesses because they naturally pursue customer delight, moats, scale and large markets. Bom Kim’s two-week focus on diaper costs and his detailed route toward “the best e-commerce experience in the world” exemplify “credible aggression”; Greenoaks ultimately invested just under $1 billion in Coupang over ten years and continued buying after the IPO.
Extreme growth is healthy when it is the output of product-market fit, not an input pursued for its own sake. Mehta wants growth endurance, considers some organizational breakage acceptable and calls rapid expansion a “moral obligation” when fit is strong: a disproportionate share of Wiz’s go-to-market and engineering staff served their country, yet Wiz produced one of its best quarters; Coupang rebuilt its growth culture after a year of 18% growth.
Industrialized venture capital may fund more companies, but Mehta argues it creates less competition for the truly exceptional ones. Sector-stage-geography matrices optimize for coverage and can compress returns from historic 35% net IRRs toward the teens, yet two Series B companies with $30 million of ARR and 100% growth can receive similar terms while one becomes worth billions and the other approaches insolvency. “What you’re sacrificing is fidelity and insight.”
A prepared mind lets Greenoaks deploy hundreds of millions during volatility rather than waiting for consensus. Navan’s revenue fell from $100 million to zero during COVID, but a roughly four-day financing arrangement helped it move from perhaps No. 4–8 to the industry’s top two; during the SVB weekend, Greenoaks agreed in about 30 minutes to provide Rippling $500 million so every customer’s payroll arrived Monday.
Carvana captures both the upside and pain of concentration: Greenoaks bought from roughly $50 down toward $5 while the company lost about $5,000 per vehicle including interest. The thesis was that Ernie could reverse unit economics, stabilize reflexive debt and eventually sell 2–5 million of roughly 40 million annual used-car units; instead of placating markets with immediate cuts, he slowed down and A/B-tested what the business could remove without destroying its future.
AI does not suspend the “laws of great businesses”: customer delight, broken trade-offs, defensibility and a large market still govern. Mehta says DeepSeek R1 achieved a 35× reduction for input/output tokens relative to OpenAI reasoning models, but he works backward to customer consequences and remains skeptical of model companies requiring huge capital reinvestment every 12 months amid rapid catch-up—while conceding that valuations have so far made him look wrong and ChatGPT proved a consumer layer can work.

Deep dive

1. Craftsmanship became Mehta’s template for recognizing exceptional companies

Mehta’s Jain grandfather, Dharichand, meditated for an hour or more daily yet owned a Mumbai gun shop. He discussed each weapon less as machinery than as art: the design and craftsmanship of every component mattered more than “what the capability of that gun would be.”
That childhood translated directly into Greenoaks’ metaphor of founders as painters. If investors evaluate painters, Mehta tells younger colleagues, they must study the types of tapestry they could use, the paints and different forms—then spend time with artists themselves.
The resulting standard is comparative: repeated exposure teaches an investor what exceptional quality looks like against work that is merely competent. Greenoaks’ broader motivation, Mehta says, is an attraction to “beautiful businesses” and “beautiful relationships.”

2. Jaw-dropping experiences begin where ordinary companies accept trade-offs

JDCE—“jaw-dropping customer experience”—is so embedded at Greenoaks that Mehta jokes the team would discuss it all night after five drinks: “It’s like we have it tattooed on our arm.”
The mechanism starts with breaking a trade-off through something technically or operationally “borderline impossible.” Customers often cannot articulate every pain point; they can only describe frustration, leaving the founder to identify and solve problems competitors have normalized.
Coupang created the term for Mehta. Koreans initially said two-and-a-half-to-four-day delivery was fine, but Bom Kim pursued consistent delivery within 12–24 hours. That required new warehouse-management and routing software, football-field-sized warehouses, local delivery camps, apartment-level distribution, specialized packaging and rules for safely delivering at 6 a.m.
The early unit economics broke, inventory purchasing was difficult and the flywheel took two to four years. Yet retention moved from market levels in the 30s to Rocket cohorts in the 60s; interviewed mothers cried over diapers arriving by morning and pleaded, “Please don’t take this away from me.” Mehta’s verdict: “That is not an NPS score of nine.”

3. Coupang joined founder judgment, business-model work and decade-long concentration

O’Shaughnessy’s introduction says Greenoaks put 40% of its initial $50 million fund into Coupang, eventually producing roughly $8 billion. Mehta says the firm led five rounds, invested just under $1 billion across ten years and bought additional shares in two or three post-IPO years.
Bom’s defining trait was literal focus. His calendar might contain nothing but diaper cost-of-goods negotiations for two weeks, five or six hours daily, while everything else burned. “Focus means saying no to everything else”—after correctly identifying the single most important task.
Ambition became credible because Bom could map the mountain: route, supplies, trade-offs, team and failed paths. He wanted the world’s best e-commerce experience, not merely Korea’s, and backed that claim with 24-hour cycle times, beds in delivery camps and near-continuous conversations with Mehta.
O’Shaughnessy pressed whether Greenoaks was founder-centric or model-centric. Mehta’s answer: it always starts with founders, because he has encountered perhaps only one or two extraordinary founders building bad businesses—and they learned quickly. Deep model work then helps Greenoaks distinguish temporary volatility from thesis failure: “Good businesses are hidden in bad P&Ls.”

4. Durable growth matters more than quarterly smoothness

Mehta keeps the S&P 500 list on his desk and asks what missing names will eventually join it. His estimate is that 1% of constituents create 90% of the value by pulling free cash flow from legacy incumbents over decades; quarterly Netflix debates matter far less than whether Netflix can compound for 20–30 years.
O’Shaughnessy challenged the hidden cost of speed. Mehta conceded that “growth is an output, not an input,” but argued that the best technology companies exhibit growth persistence, with the following year’s growth potentially remaining in the 80–90% range of the prior year’s. For bits businesses especially, “unreasonably high growth” is healthy, even when some things break.
When Hamas attacked Israel and a disproportionate share of Wiz’s go-to-market and engineering employees left to serve, Greenoaks expected November and December to be written off; Wiz instead produced one of its best quarters. After periods when stockouts, drivers and warehouse capacity constrained what Coupang could offer, the company had a year of 18% growth, and advisers suggested merely returning to 30%. Rebuilding a high-growth culture above that level was, Bom later told Mehta, “one of the hardest things we did” and one of the most important.

5. The best first meeting should feel like the sixth

Selection begins before meeting anyone. Among tens of thousands of market meetings, Mehta estimates perhaps 200 annually where Greenoaks could be differentially useful—and only 10–15 founders warrant the firm’s deepest pursuit. Preparation goes far beyond testing a website; the opening conversation should feel like meeting five or six.
Mehta prefers visiting the company, even though travel reduces meeting volume. His recurring test is: if every employee were asked whether the company’s best days are ahead or behind, what share—especially among essential people—would choose ahead? That energy can matter more initially than growth, margins or moat terminology.
The founder archetype combines focus, ambition, determination and divergent thinking: seeing something others would “vehemently disagree with” and being right. Greenoaks then looks for JDCE and defenses such as network effects, shared-skill economies, counter-positioning or hoarded resources. Mehta calls the method simple enough to sell “40 to 50 points of IQ”; the edge is disciplined repetition.

6. Venture’s coverage machine creates capital abundance but insight scarcity

Mehta sees the industry’s shift from six Sand Hill firms to a scaled asset class as economically logical. Historic 35% net IRRs were too high to persist; deploying more capital through more people can lower returns into the teens while still satisfying investors.
The dominant model is a sector-stage-geography matrix measuring what percentage of each quarter’s rounds the firm covered. Mehta calls this the “private equitization” of venture and thinks it is appropriate for more than 90% of the market—but precisely wrong for the 10–15 best founders, who should receive the entire firm rather than one box in its matrix.
O’Shaughnessy’s pushback was supply and demand: he offered an illustrative comparison of “$2 billion when Don Valentine did NVIDIA” with “$200 billion or something like that” today. Mehta held both propositions as true—too much money funds thousands of companies, yet specialization and institutional layering leave fewer firms able to pursue exceptional founders with concentrated judgment.
His evidence is pricing dispersion that fails to reflect outcome dispersion. Two Series B companies can each have $30 million of ARR, grow 100% and trade on nearly identical terms; five years later, one may be worth many billions and the other nearly insolvent. Founders also choose on partner, brand, speed and understanding—not valuation alone—so coverage sacrifices “fidelity and insight.”

7. Greenoaks replaced exhaustive coverage with faster, narrower conviction

The firm’s information density is powered less by secret machinery than lifestyle. Mehta says the team has “no beach houses,” likes spending 80 hours a week together, is not on Twitter and would rather study companies than attend basketball games.
Two or three years earlier, individuals could take 12–30 meetings weekly, and Greenoaks proudly showed investors 92% Series B coverage. Mehta now regards that as the wrong optimization: the largest improvement has been becoming “much better, not a little better” at separating the vital few from the trivial many.
The refined flywheel combines faster information asymmetry with differentiated insight about long-term enterprise value. Feedback reasonably asks what Greenoaks has done lately, challenges apparently expensive rounds and questions its intensity, hiring misses and fast firing. Mehta says the evidence may take ten years to surface, while acknowledging that the firm could be wrong on some high-priced rounds.

8. Outsourcing judgment produced a major mistake

Early Greenoaks heard that Elon Musk fired quickly, micromanaged deeply, disappeared and returned to change everything. Mentors concluded he was “not backable,” and the firm failed to do its own primary work on Musk. Mehta calls that “the biggest mistake we’ve ever made at Greenoaks”: negative references read worse than the underlying reality.
The inversion now informs sourcing. Greenoaks likes founders in the weeds, micromanagers and people willing to fire fast; divergent founders may reject their team’s consensus because they possess contrary evidence and the determination to act on it.
Mehta’s partnership with Benny Perez supplies an internal check. They met in college over an obscure cruise-ship financing structure; Mehta still considers Benny’s “zero-to-one clock speed” the fastest he has seen and his ability to restore the firm’s long time horizon invaluable during the fog of war. Team debates can last four hours, followed by Mehta and Benny talking from 9 p.m. to 1 a.m.

9. Tencent converted a special-situations investor into a growth investor

In Hong Kong around 2007, Mehta was meant to study real estate and distressed assets for D. E. Shaw. At the Beijing Olympics, he noticed spectators increasingly using phones and learned about QQ, a predecessor to WeChat that he says was adding roughly 30 million subscribers monthly.
The comparison reset his priorities: D. E. Shaw wanted analysis of 150-year-old distressed European banks with 30 million total depositors, while Tencent was adding that many users every month. “I wanna spend all my time on Tencent,” he concluded, eventually returning to San Francisco to start Greenoaks.
The first $50 million fund was raised without a seed deal or committed backer. Mehta and Perez shared a DoubleTree room and borrowed Blackstone friends’ printers for pitch decks. Henry Kravis listened despite their missing the dress code, committed immediately, supplied introductions and later visited their office—an act of validation Kravis apparently repeated for many emerging managers.
Rather than ask investors to trust raw intelligence, Mehta pitched actual ideas: Palantir, Flipkart, Oyo Rooms and Coupang, supported by team analysis, business quality and returns math. The thesis was a 20-to-30-year technology and internet opportunity. He deleted personal email because “this is what I wanted to do for the rest of my life.”

10. AI changes capabilities, not the laws of business

Mehta compares AI investing with the Wright brothers recognizing that flight still obeyed aerodynamics. Great companies likewise must delight customers, break hard trade-offs, defend themselves and address large markets. Greenoaks’ crypto mistakes came when it accepted that boards, auditors and customers somehow no longer mattered: “The laws are different this time. They never are.”
DeepSeek R1’s roughly 35× reduction for input/output tokens versus OpenAI reasoning models impressed him, but the investment question is not the benchmark alone. Greenoaks raises the abstraction level to ask what the capability changes for customers, then works backward.
Mehta remains skeptical of model businesses that require enormous capex, receive a payoff and must reinvest 12 months later as competitors rapidly catch up. He explicitly marks uncertainty: rising valuations have made that view look wrong, while ChatGPT demonstrated that a large consumer business can sit atop a model.

11. Navan and Rippling show what committee-free capital can buy

When COVID took TripActions—now Navan—from $100 million of revenue to zero overnight, Mehta called within a week. Convinced it remained the market’s end-state solution, he offered up to $500 million so it could capture share rather than retreat; the arrangement took about four days.
Over the next two years, Navan moved from perhaps fourth, fifth or even eighth in travel management to the top two. The financing expressed Greenoaks’ preference for using volatility when a prepared view of the end state remains intact.
During the SVB weekend, Rippling faced a plumbing failure, not its own insolvency: SVB pooled customer payroll funds. Parker called Friday because delayed employee payments were unacceptable. Greenoaks agreed in roughly 30 minutes to invest $500 million; after 48 hours of continuous work and despite signs Sunday that things might be okay, Parker proceeded with the deal. Every Rippling customer was paid Monday.

12. Carvana turned a near-collapse into a test of founder temperament

Greenoaks nearly invested about $500 million when Carvana fell into the $30s during COVID, then withdrew for internal reasons and watched shares approach $300. After the ADESA acquisition added debt and used-car conditions reversed, market value fell from roughly $70 billion to $1 billion—an apparently unprecedented non-fraud collapse of that scale.
The firm began buying around $50 and continued toward $5. Carvana was losing about $3,000 per unit on an EBITDA basis plus $2,000 of interest, making bankruptcy look like a question of when. As partner Ben put it, “What’s the difference between being down 95% and 97.5%?” Answer: half the remaining capital.
The two-part thesis was that Ernie had enough runway to reverse operations, after which the reflexive debt problem would improve; then Carvana’s differentiated buying experience could scale from 400,000–500,000 units toward 2–5 million in a roughly 40-million-unit annual market.
Greenoaks prescribed immediate line-item cuts, but Ernie instead ran A/B tests to learn what could be removed while preserving the ability to manage through the crisis and grow afterward. At dinner amid hostile coverage, he worried about employees’ children hearing the company would fail, not his own reputation. Even searching ten minutes for a Marriott gift certificate reinforced for Mehta how deliberately Ernie could slow down while markets panicked.

13. Fund size follows the largest useful check, not the largest possible AUM

Mehta’s choice is “the hall of fame of returns or the hall of fame of AUM.” Greenoaks wants enough capital to answer the few annual calls requesting $500 million to more than $1 billion, but not enough to force weaker investments.
Funds grew from tens of millions into billions while company count fell toward 10–12, versus a historical high near 15. Across approximately $15 billion of AUM, Greenoaks owns only 55 companies; Mehta says he could contact every founder in half a day.
He assigns zero value to Greenoaks’ saleable enterprise value—“I have no plans to sell my painting”—and keeps outside LPs because he enjoys them, values the courage they provide and wants Greenoaks to rank among their best-returning investments.

14. The alternative-capital experiment failed because frontier insurance was not merely cheap

Greenoaks originally paired a normal fund with GGH, a $150 million holding company intended to buy 51–100% of frontier-market insurers. The thesis used an S-curve linking GDP per capita to insurance penetration: backing the best P&C carrier could be a levered claim on national development, while premiums supplied investable float.
Reality supplied different risks. Before Mehta’s first Nigerian trip, his banker described a London visitor finding a corpse in his hotel bed, becoming a police suspect and needing roughly $10,000 to get released. Greenoaks nevertheless bought insurers in Nigeria and Pakistan; Mehta spoke especially highly of the Pakistani partner and experience.
Rwanda exposed the model’s weakness. Its leading P&C insurer lost $1.80 for every $1 of premium, while unaudited rivals survived by continually writing more business. In a death claim, the supposedly deceased claimant appeared in court and the proceeding continued anyway. Mehta concluded, “There’s no winning here.”
The project consumed real money, years and attention, and Mehta calls it “the single biggest mistake we’ve made at Greenoaks, I think.” It clarified the desired combination of founders, markets and institutional conditions while abandoning the insurance-company route to an alternative capital structure.

15. Fillmore Street applies the same customer-delight thesis without financial return

Mehta created a nonprofit with Cody Allen to revive several blocks of Fillmore Street in San Francisco. The economics are intentionally bad: buildings bought near a 5.25% cap rate when Treasuries yielded the same, followed by tenant improvements for mom-and-pop restaurants at roughly a 3% cap that barely covers rent.
His premise is that San Francisco uniquely attracts people who build their version of the future, yet anti-business, anti-growth, high-tax and anti-family policies could disperse that cluster. Budapest’s prewar physicists are his warning that a concentration of talent can disappear and cannot be assumed to return.
Former politician Aaron Peskin used picket signs bearing Mehta’s face and accused a billionaire of taking over the city without first contacting him. Mehta says the actual work is preservation: coffee shops, an all-day diner, restaurants and a rebuilt theater, with local entrepreneurs creating experiences that delight people.

16. Mehta admires bold investors but protects his freedom to change his mind

His controversial choice for investing’s greatest is Yuri Milner. Milner’s Facebook investment in a money-losing company at roughly $10 billion in enterprise value shattered Mehta’s mental model of permissible growth investing; he admires DST’s ratio of correct decisions to impairment and category-defining bets such as ByteDance and Xiaomi.
Mehta also defends Masa, whose Arm critics supplied ten technical reasons the deal would fail. Masa’s answer—“He fails to realize that the market’s growing”—proved decisive. Mehta further values investors by whether they support entrepreneurs when conditions deteriorate, something he has watched Masa do repeatedly.
Mike Moritz supplied a quieter model: during political attacks over Fillmore Street, he encouraged Mehta to write a transparent op-ed and personally helped him despite having nothing to gain.
Public attention still worries Mehta because written opinions become identities that demand defense. He prefers trying on a view “like a sport coat” and discarding it without pride of authorship. The final lesson came from school mentor Joe Rosenthal, who stopped his showboating after a soccer goal and reminded him that teammates and coaches made it possible: “Have some class.”

Random Ramblings April 2025 Market Overview, April 15, 2025