Summary

Horing’s answer to the $100B Vision Fund question is venture buyouts at scale: the cleanest private-equity example of a return was VMware — EMC paid ~$650M and sold for $60B — and Insight has repeatedly made 5-6x taking control of smaller software companies. His dream was competing with Microsoft or Palo Alto for a founder’s company: “I could sell to Jeff for a billion dollars my next Instagram and still retain massive ownership.” Masa did exactly one deal like that — ARM, roughly 4x on a huge check.
On fund size, Insight is “pretty close” to optimal at ~$12B, deploying $3B+/year — not capital constrained, and deliberately not chasing IPO-plus-plus late-stage rounds where “the spreadsheets start to look like two to 3x.” The neglected middle — unlevered 30-40% growers, “tweeners” too slow for minority VCs and too small-TAM for strategics — is where hit rates are “super high” and competitors few.
The metrics call: gross retention (GDR), not net retention, drives all exit value. Low-80s GDR at $1B revenue means refilling $200M of churn every year before growing; the “second derivative” framing — change in new business — separates Google-like compounders from vertical-software saturation traps. Wiz doubled or more its net new bookings every year for six years. “Buying cheap in technology… there’s not a long list of really rich people who’ve done that.”
The one-fund structure is defended on portfolio theory grounds: check size, not fund charter, manages risk, and “the best bet on the table is the double down bet” — an 11 against a five. Insight’s follow-ons were its best checks, including a $5M position ridden to a $1B position; separate growth funds create tweener conflicts and “am I bailing the company out?” optics. Cost: “you lose a little bit of discipline from third party pricing.”
Post-2021 marks may not reflect reality: secondaries trading at 70 cents on the dollar say so, and low-single-digit growth with 80% GDR is worth maybe 3-4x revenue — the margin math (20% margin business at 15x cash flow = 3x revenue; 50% margin = 7.5x) eventually catches everyone.
On AI, the bear case on software (“I’ll just use Claude to build my next SAP”) doesn’t cost him sleep: “It was never a technology barrier. It was always a business knowledge barrier.” The real risk is budget shifting from software to AI; the real prize is TAM expansion — automating the 95% of a user’s day spent generating data — with ~25 agentic AI bets in the portfolio and Anthropic held via the public strategy, not the core fund.
Firm-building thesis: investing reduces to four jobs — find, win, select, make them work — and Insight pursued scale inside software rather than by check size, geography (the 2000 European fund was “brutally painful… it’s really hard to export judgment”), or new asset classes. The sourcing machine — 60-80 engaged people with claimed lanes — has produced 16-18 alum-founded funds and 30+ partners at other firms.

Deep dive

1. The $100B thought experiment: venture buyouts, not late-stage growth

Patrick opens by handing Horing the SoftBank Vision Fund and asking how he’d deploy $100B. Horing goes backwards to the best PE/venture deals ever: VMware is the cleanest — EMC spent ~$650M, sold for $60B; Instagram and YouTube each roughly a billion to a trillion, and “a lot of that was going to happen independent” of the acquirers — PayPal had “almost no real eBay effect.”
His dream strategy: compete with Microsoft, Palo Alto, or eBay for control of growth-stage software companies, where the founder “could have my cake and eat it too… sell to Jeff for a billion dollars my next Instagram and still retain massive ownership.” Insight’s venture buyouts at ~$100M checks made 5-6x+; Masa’s one deal in this mold was ARM, ~4x on a really big check.
Today it’s tougher: “to get Benchmark or Sequoia to sell something for a billion dollars probably doesn’t get their heart rate up.” Sub-$1B remains the sweet spot, but strategic buyers distort it — “Palo Alto pays $500 million for 30 guys in Israel, that’s a different game.”

2. $12B is enough; late-stage privates rarely pencil

Asked to size the optimal fund, Horing says “I think we’re pretty close” at ~$12B, deploying $3B+ a year: “we definitely don’t feel capital constrained to the opportunity set.” Buying OpenAI or Anthropic in volume at latest-stage prices would be “a better version of that Vision Fund” — but “we don’t lose sleep over that not being our core strategy.”
On the Databricks-scale private market: companies are staying private through what are effectively IPO-plus-plus rounds. His filter is unchanged — forecast, likely exit value, return on capital — and only “once in a very rare while” does something at size price for risk-adjusted venture-like returns.

3. From 11th-grade to college math on software metrics

Six years ago Horing told LPs Insight was at “11th grade math on software” while the industry was at seventh grade; “we’re getting closer to college today.” His example: when Wall Street was obsessed with net retention at one of his IPOs, he called it “one of the least informative numbers I could think of” — GDR is what matters: low-80s gross retention at $1B revenue means finding $200M of new business annually just to fill the bucket, which shows up directly in CAC.
The “second derivative” insight (he considered naming the hedge fund after it): the change in new business beats the growth rate. Companies with flat new-bookings adds can still print 75-100% growth off a small base — “you can copyright that number if it’s flat” — common in vertical software where you saturate the annual decisions; model five years flat and your exit growth rate looks nothing like Google’s 15-year 100% compounding.
Everything zooms out to LTV/CAC: GDR predicts LTV; then the question is market pull — “if I added 10 million of new business this year, can I add 20 million next year and 40 the year after?” Wiz doubled or more its net new bookings each year for six years. Half his partners refuse to compromise on these metrics at all: nine out of ten times they’ve driven Insight’s wins, and “we will not do a low gross retention business today unless we really are confident we could change it.”

4. The five ingredients of the perfect investment

Horing reads from his cheat note: big ROI, big ASP, time to value, CEO, and tech team. Wiz “checked all five”; Monday most of them. “Getting the time to value and the ASP is really rare.”
His TAM heuristic runs through selling price, not top-down multiplication: a likely agentic-AI company selling $500K/year into hospitals against Epic’s $10M average is 1/20th the size at best — “best case I’m probably 1/20th the size of Epic.” Top-down TAM math “is riddled with errors.”
Time to value is his hidden data point: pointing a cursor at OpenAI versus a three-year SAP implementation journey — the latter inherently caps how fast decisions get made. And for those sticky, long-install businesses, ~15 referenceable customers is the magic number for inflecting growth: below it you’re missionary selling and can’t scale sales; above it “you have supply constraints to scaling, not demand constraints.”
Honest hedging on the CEO ingredient: “I sometimes write that story after the fact… I certainly have plenty of phenomenal CEOs that were rejected by a lot of other firms.”

5. One fund: risk by check size, and the double-down bet

The strategy traces to Warburg Pincus (stage- and industry-agnostic since 1968). Two classic-portfolio-theory advantages: first, risk management via check size — the Vision Fund could write a $200M check that was inconsequential to fund returns; in a $1B fund that’s the $2M flyer “I’m not going to get too worked up over.”
Second: “the best bet on the table is the double down bet… you’ve got an 11 against a five.” Insight’s double-down checks were its best checks — a $5M position grown to a $1B position (“we never would have seen the billion dollar position without the relationship”), and Monday and other large exits that started under $25M and grew to $200M through secondaries and follow-ons. Separate growth funds breed tweener conflicts and cross-fund optics — “am I bailing the company out?… it’s not zero.”
The downsides, unprompted: “you lose a little bit of discipline from third party pricing… you could kind of believe your own BS,” and the $2M kick-save checks risk “chasing good money after bad” — though plenty of bridges to sale have actually worked. To LPs, “we don’t fit into a bucket. My whole life I’ve never fit into a clean bucket.”
The founding trauma behind it: watching top firms impose $50M/$100M check minimums as they scaled — “I don’t want capital to dictate my strategy.” And the late-stage pivot has become a bit of a necessity: pre-IPO rounds that once returned 4-5x now model at 2-3x, so getting on the balance sheet earlier “is a bit of a necessity.”

6. Origin story: Kevin Landry’s playbook and software outside the Valley

Nobody would hire Horing to do small software deals in the early ’90s — software then was “IBM and Microsoft… number 50 was like $10 million.” At a conference, TA Associates’ Kevin Landry laid out his sourcing playbook — ripping help-wanted ads out of the New York Times and cold-calling companies — with a “here it is, I don’t really care, good luck.” Horing, 26, started doing it at Warburg and “realized I could find deals all day long like this.”
The structural insight: application software clustered near customers, not Silicon Valley — banking software in New York, pharma apps in New Jersey — and most software companies were bootstrapped consulting projects findable “well after their incubation phase,” counter to the West Coast model. Insight launched on a $16M blind pool from high-net-worth individuals, having sourced deals before it even had a fund.

7. The sourcing machine: 60-80 engaged people, claimed lanes, a coaching tree

Chapters: Horing and his partner dialing; then Mike Triplett from Summit; then the 1999 decision to hire undergrads directly — Goldman/McKinsey alumni “get spoiled and don’t want to go back to picking up a phone.” Today it’s 60-80 people, run like a software sales org: claim a deal onto your pipe, rules on how long before it’s “up for grabs again,” comp deliberately over-rewarding cooperation because “we’re all comped the same… one for all, all for one.”
The analysts run the partners’ calendars — “every meeting today was set up by the analysts” — and Horing has flown to Estonia, Sweden, and five other places this year as analysts set up meetings. The trade-off he flags: partners at other firms get a natural deal-flow regulator from busy calendars; Insight’s systematized meetings “can go on forever.”
What makes a great sourcer: “classic sales skills — hunger, winning, probably lack of self-awareness… ability to handle rejection, combined with a lot of content.” His favorite proof: a CEO whose first question in their meeting was “where’s the analyst?” — the relationships get that deep. The candid downside: some founders are turned off by being called by junior staff, and mass outreach means mass rejection, which Insight tries to frame as “postpone, not reject.”
The output is a Bill Parcells coaching tree: 16-18 funds founded by Insight alumni, 30+ partners at other firms, nearly every current IC-level partner homegrown. Why it works: “the training you get at 23 at Insight is like no other job… all you’re seeing is at-bats. Investing is pattern recognition — you could be the smartest guy in the world, but if you don’t see the deals, your track record is not going to be that good.”

8. “Stage is not a strategy”: the venture-buyout franchise

In the ’90s, “giving cash to a founder was like a four-letter word” — secondary sales taboo, TA and Summit just starting to break it. Insight’s confidence grew deal by deal: buying out a departed founder, buying a software company from a large insurer (now called, per Horing, Veraphor — one of the earliest software buyouts, “high-fiving with two times leverage from a crazy hedge fund”; today it carries ~9x), and taking control of an Australian company whose “founders were ready to go surfing” with a CEO in Insight’s pocket.
The core hunting ground: “tweeners” — unlevered 30-40% growers too slow for minority investors except at big discounts, too small-TAM for strategics — “my suspicion is there’s tens of thousands of those now.” His line to LPs: “Stage is not a strategy to me. Everything else is just a spreadsheet… risk-adjust it and put a price against it.”
Current fund mix: ~10% early stage, ~30% growth, ~30-40% venture/growth buyout, ~20% LBOs, deliberately fluid with capital flows. Return targets: a ~30% gross target for most, with buyouts allowed five points lower — with “super high” hit rates in venture buyouts because the sourcing engine finds “salmon software to the salmon industry… people jumping on planes and calling up companies in Norway.”
On Anthropic: “that’s a little bit more driven from our public strategy, not the core fund.” A change of mind admitted: “had I heard the story two years ago I would have had a much more positive view earlier” — Darius is “phenomenal… the second you hear, you’re like okay, there is a moat or there might be a moat,” and the coding inflection of the last four months made the numbers “incredible.”

9. The Madden ratings: see everything, pick in the middle, win by showing up

Asked to score himself on see/pick/win, Horing gives the firm a “crazy high” sourcing score — “we see every pitch” — and says “picking in the middle is awesome; picking on the edge is okay.” On early stage, “the whisper is so strong” in California: “those guys are awesome… they’re using different dots to make what I’d call intangibles work.”
Winning is machinery plus hustle: “showing up hard, getting on planes, invited or uninvited,” backed by super-smart McKinsey operators, a large interest in Riviera Partners (the largest tech recruiting firm for CTO/CPO talent — “in the AI age talent is everything”), and 15 people doing nothing but Fortune 500 introductions. A CEO’s two asks are revenue and people; Insight built around both, sometimes delivering a company’s first $10M of revenue.
His summary of the whole job, told to LPs on day one: “You got to find deals. You got to win deals. You got to pick them and you got to make them work. That’s it. That’s my job.”

10. Scaling judgment: from tennis to soccer

On firm strategy, Horing cites a likely Marc Andreessen quote — “I’m a marketing firm with an investment arm” — and contrasts it with his own: “I wanted to be a software company with an investment arm — outsource software know-how to us, with an investment arm to monetize it.” Horing says the majority of firms he heard on the podcast struck him as having really good strategies and being really good investors.
The 2015 scale debate: LPs had “an allergic reaction to the word scale” even though sourcing, winning, and operations all clearly improve with it. The scary scaling paths — bigger checks (differently priced, differently competitive), geography (“oh was that painful… it’s really hard to export judgment” — the 2000 European fund nearly made him pull his hair out; Insight is ~90% New York by headcount, IC entirely so), and new asset classes. Instead: same underwriting criteria, growing with a market that got bigger — “Databricks, if you just divide everything by 10, looks like a great classic growth deal… OpenAI, turn 12 billion into 12 million and you’re like, wow, why wouldn’t I write a $10 million check?”
Selection was the hard part: the “tennis match” model — young partner pitches a half-listening Jeff, deal blows up, “young partner is now stuck” — didn’t work. The fix: about eight teams of IC members (six together 25+ years), the senior partner meeting every company and owning every deal — “if somebody screws up, it’s on me and there’s no hiding it.”
Concept deals — big-price bets on team and technology without numbers — exist but only Horing does them, small and check-size-managed: “even I’m not doing that right now.”

11. Marks can be wrong: the post-2021 valuation math

The 2021 cohort doesn’t necessarily have to die because it was overfunded, and marks may not have caught reality: “go look at how some things trade in the secondary market… when you’re at 70 cents on the dollar, your marks aren’t right.” For low-single-digit growth with 80% GDR, fair value could be “a really disappointing three or four times revenues.”
The mechanism: 12-month CAC plus four-year customer life present-values to maybe 2-2.5x per dollar invested before R&D and G&A — a 10-20% margin business — versus 50-60% margins for 100% GDR companies. At 15x cash flow, that’s 3x revenue versus 7.5x. Exceptions exist (a mid-80s-GDR company with 3-month CAC that Vista now owns), but “all that matters is multiples of cash flow and predictability of that cash flow in a recession” — core banking systems don’t get ripped out.

12. AI: not an existential threat to software, a TAM accelerator

Horing’s credibility check on skepticism: he sat out blockchain — “it’s 12 years in, come on” — with the killer anecdote that “all the technology guys think the tech is kind of meh but the finance aspects are cool, and all the finance guys think the finance isn’t so great but the tech looks really cool. Nobody who understood database technology said this is the best database I’ve ever seen.” Vision AI, which he backed pre-ChatGPT, “never got the buzz… I can’t explain it” — but language exploded, and watching his own family convert from Google to Gemini or ChatGPT in three months was “religious.”
The bear case — “I could just use Claude to build my next SAP” — gets a categorical dismissal: “That’s not what software ever was. It was never a technology barrier. It was always a business knowledge barrier.” Development costs are “inching down, not collapsing overnight — and I can’t explain exactly why.” The 4GL analogy: a profound productivity tool that didn’t displace SAP.
The real risk he does concede: budget migration — “that’s just not the cool kid on the block to buy a CRM today… that matters, that’s growth rates.” The upside: massive TAM expansion — in vertical CRM-like apps, “95% of the person’s day is generating and getting the data; if I could automate a big portion of that, that’s hugely valuable.” Half a dozen portfolio companies are re-accelerating on AI products; Adobe could go from “3% market share of humans to Photoshop” to 20% as the learning curve collapses. Insight holds ~25 agentic AI bets it thinks “could be really profound.”

13. The tush push: X factor, forgiveness, and defending risk appetite

Career evaluation in a business of long, lucky outputs uses his “X factor”: “if I took you out, what would have happened? Would we have sourced the deal, won it, decided to do it?” One of his best partners “had a really slow start… made a lot of mistakes, but I saw his inputs were great — now he’s probably the best investor in the firm.”
Two archetypes of senior partner: one holds you back from the cliff, “the other one’s shoving you over the cliff… I’ve got your back if it doesn’t work out.” Horing is emphatically the latter — “I call it the tush push” — because “if you look at generational firms, risk appetite is probably the biggest challenge… if you don’t get fired you’re going to be pretty successful, so the impetus to stick your neck out is really low.” His biggest frustration with a departed partner was “the things he didn’t do… he always had five reasons not to do it.”
On drive: he’s competitive against his own score, not other people — “if you shoot a 65, I’m high-fiving you, I’ll buy you a beer… I just want to get my own score as low as possible” — and thinks the firm’s culture self-selected for that.

14. Selling windows, honest misses, and a more boring decade ahead

On selling: “the easy things come naturally — IPO, strategic knocks, you sell. The harder one is when you have to push it.” The scar: the ‘99 fund had a 4x in public markets it couldn’t sell through lockup; by expiry it was a 1x. COVID-era errors owned openly: virtual-conference bets that “had no legs after COVID,” and “decision-making probably not what we thought it was over Zoom… a year of remote work — really, really bad. Never going to do that again.”
The misses cataloged without spin: Insight passed on Uber “at a really attractive round — we fought like hell as a partnership… obviously huge mistake,” exited Twitter before a likely Musk takeover, and largely skipped consumer internet and mobile — “our misses are so high in those categories that we’re like, whatever. We can’t be everybody to everything.”
Next decade: “much more of a rinse and repeat model… a little bit more boring,” with the bar “never higher” since summer 2022. The cultural essentials: “we don’t have to be the loudest voice at the table, ever — we want to be the most helpful voice, and we don’t need credit,” plus never giving up on sideways deals — “those are the worst hours of ROI you can possibly get, but it feels like the right thing to do.”
The closing kindness question: the Warburg partner who “pulled me out of a hat” after ~100 rejections, and Steve Freeman, the just-retired Goldman Sachs CEO, who “for reasons I still don’t know, took me under” and later brought in Bob Rubin as a second mentor in Insight’s first decade.

The Startup Powering The Data Behind AGI Tulsi Gabbard on Russiagate Hoax Evidence and How She’s Reforming Politicized Intelligence Agencies