The Playbook on Buying and Running Companies Forever

Summary

• Bending Spoons is “25% private equity, 75% tech company”: it buys digital businesses outright, off its own balance sheet, “to own and operate forever,” then rebuilds them root-and-branch — code, cloud, UI, monetization, org. The result: ~1.3B revenue in 2025 with per-share revenue or EBITDA growth still compounding ~75% a year over the past four years, capped by a $700M equity raise at an $11B valuation — the largest round by any private company in Italian history.
• The founding insight came from a failed startup: luck plays a huge role in zero-to-one — Ferrari watched a couple dozen peer teams and saw “almost no correlation” between talent and who came out on top — while one-to-N excellence compounds with discipline. So buy from people who got lucky or whose passions changed, and run the asset better. “We don’t want to bet our entrepreneurial lives on getting lucky.”
• The moat is the platform: fluid R&D across units (most companies are “years behind in terms of the optimal staffing”) plus a talent machine — 800,000 unique applications for 250 hires in 2025 — that Ferrari says would take a copycat “easily seven or eight” years to replicate even with $1B and full knowledge of the playbook. “I’ve seen nobody try cuz they just understand it’s just too painful.”
• Proof points: Evernote got ~250 significant product improvements in 2.5 years with a smaller team, sync times cut to 10% or even 1% of before, prices ~60% higher — and retention at an all-time high. AOL is, Ferrari believes, “the fifth most used email inbox in the Western world” hiding behind a legacy label. “If Google ever wants to divest it, I will be happy to take a look.”
• Deal discipline is anti-bias by construction: assumptions are debated without ever looking at model output (“that’s forbidden”), then a Monte Carlo IRR/NPV distribution becomes “the truth.” Offers are near-max, Buffett-style — Bending Spoons has never lost a bid, which Ferrari reads self-critically as evidence “we’re probably not very good at negotiating.”
• On AI eating SaaS: mostly a tailwind for a diversified consolidator whose most units are 20% of revenue or less; telling ChatGPT “build me Jira” is “not months away, it’s not even a couple of years away” — something that works the same 95% of the time is “infinitely easier” than something that works essentially 100% of the time.
• Org-design heresy that works: no variable pay, no stock grants, no KPI bonuses — fixed cash salaries with optional discounted equity purchase, because incentive plans are “absolutely guaranteed to create at least some perverse incentives” and make problem-solving transactional.

Deep dive

1. 25% private equity, 75% tech company — buy forever, rebuild everything

• Ferrari’s self-description: a “pretty unusual beast” that acquires companies as its key growth engine — 100% acquisitions, no minorities, bought off the balance sheet “to own and operate forever.” Unlike PE’s “relatively shallow interventions,” Bending Spoons will “rethink the entire company”: rewrite the software, re-architect the cloud, redesign the UI, optimize monetization, rebuild big chunks — sometimes the entirety — of the organization. Value created is reinvested into the platform to go after bigger acquisitions.
• The compounding record, in his own words: ~1.3B revenue this year, per-share revenue or EBITDA growth still ~75% per year over the past four years. “10 years go by and you look back and you go, ‘Oh, wow, I remember we were making half a million a year, now we’re making a billion.’”
• The ambition is institutional — “the company being our product,” a Berkshire Hathaway-type “defining company of its generation” — with a deliberately European seed: a 700-million-person continent with “not a single” trillion-dollar company. “We are not a periphery of the empire.” Why so few European giants? Default, not destiny: a founder is “a passionate, determined, maybe talented idiot” who never runs a locational study — more virtuous local examples would change the default.

2. Origin: a failed startup, €40,000, and a founder who cries once a decade

• Evertale (2010) was a self-writing diary of a user’s life with AI — “we were very early on AI, too early in fact.” ~€1M raised, near-bankruptcy, €40,000 left; the VC sold its shares back for a nominal €1 and suggested a nice vacation. Instead that €40k seeded Bending Spoons in 2013, all founders living in one apartment.
• The funding hack: three graduates agreed whoever landed the most lucrative job offer would work and pay rent for the other two. Ferrari got McKinsey, told the partner he’d moonlight on the startup expecting the offer withdrawn — and was encouraged instead. Patrick summed it up: “There was no contract, nothing. 100% just trust… it makes 99% of it so much more enjoyable if you don’t have to be too transactional.”
• The low point: 3-4 months of 12-16-hour days cold-emailing “anyone on the planet” for consulting work yielded one €10k contract from someone Ferrari thought took pity on them for a burger-chain app. “I cry maybe once a decade. I cried” — nearly four years of 100-hour weeks with “nothing to show for it,” supported by a more optimistic co-founder: “at least we’re in this together.”
• The name: co-founder Matteo, a Matrix fan, pitched Bending Spoons — initially rated four stars out of five in a spreadsheet, behind “App Appeal” (“thank God we didn’t pick it”). It stuck for two values: the power of the mind, and that anything worth having requires work.

3. The founding insight: luck plays a huge role in zero-to-one; one-to-N is a discipline you can own

• Watching a couple dozen peer startups, Ferrari saw “almost no correlation” between the most talented, hardest-working teams and those who came out on top — “even if you’re a genius and you work your ass off, the stars will probably not align for you anyway.” Meanwhile their functional skills — engineering, product, marketing — were on “a clear path to excellence,” a matter of perseverance and discipline. “We don’t want to bet our entrepreneurial lives on getting lucky.”
• The arbitrage: buy from people who got lucky, or whose passions changed — going from one to ten “is a different job” — creating deals that are “great for both parties.” His honest revision: the bigger edge, structural advantages of integrating businesses under one roof, was only discovered later and “today is probably more important” than the original thesis.
• Execution started absurdly small: first acquisition within a year, a €10,000 iOS keyboard-personalization app that returned €20,000, compounding 10k → 20k → 40k → 80k alongside a few scratch-built products that extended runway.

4. The platform edge: fluid R&D and a talent machine no standalone can build

• The under-appreciated inefficiency Bending Spoons attacks: R&D opportunity is fleeting but staffing is slow, so “most companies are years behind in terms of the optimal staffing.” Moving R&D and marketing resources fluidly across businesses — attacking windows, withdrawing when saturated — makes them “super efficient both on the offensive and on the defense.” (Vendor negotiation leverage adds a couple of EBITDA points — “useful, not transformative.”)
• Talent gravity: a standalone Evernote, “no matter how charismatic its leaders,” attracts “somewhat average talent”; Bending Spoons offers growth and variety, then layers on a massive AI investment predicting future performance from CVs, cover letters, and test results — unjustifiable for a company hiring 20 people a year. Net: ~800,000 unique applications in 2025 for 250 hires — one in three-to-four thousand. His Slack label is still “recruiting.”
• Ferrari’s caveat on the brand behind that funnel: “You can’t shortcut it. It takes forever — but you couldn’t as a standalone company even if you had the patience.”

5. Jobs are the product; consensus is the enemy

• “We think about the jobs we offer as our most important product.” Most companies are “too vanilla, too boring… they’re everything and nothing” — Bending Spoons explicitly targets brilliant, hungry people who want maximum talent density and are wired like Djokovic and Nadal: “they love the idea that it was supposed to be impossible.” General managers run businesses of 50-100 million in revenue at age 27-28.
• The compensation heresy: fixed cash salaries only — no variable pay, no stock grants — with an option to invest part of pay in equity at a discount. KPI plans are costly, “absolutely guaranteed to create at least some perverse incentives,” and corrode problem-solving: “it’s difficult to have a proper session where all we’re thinking about is how do we win together.” Hire high-integrity people instead, and “nine out of 10 will take it to heart.”
• His hardest personal lesson: “consensus is overrated and even dangerous.” Naturally consensus-seeking, he calls himself “absolutely terrible at it” for the first seven or eight years; once you’ve listened with intellectual honesty, “being able to just accept disagreement, pissing some people off, and going straight toward that goal is a superpower.”
• Culture rituals: “State of the Spoon,” a twice-yearly internal Apple-keynote where teams present achievements and failures with comedy (“at the end my jaw is painful cuz I laughed too much — we’re not saving lives”), and a yearly 7-9-day all-company retreat (Seychelles, Mauritius, Japan, Australia) that “pays dividends” in trust.

6. Deal discipline: never look at the output while setting assumptions

• The criteria haven’t changed since the €10k deals: digital technology (stay in the circle of competence, “here and there take a step half outside”), scale — roughly five acquisitions a year, ten at a stretch, fewer-but-bigger since effort doesn’t scale linearly with revenue (“we’re hoping to invest easily a billion plus”), predictable future performance, and the ability to make substantial improvements.
• The anti-bias machine: every assumption gets a probability distribution, debated extensively “without ever looking at what the model will spit out. That’s forbidden” — because seeing the P&L invites “maybe this is conservative, let me push it.” Only then a Monte Carlo produces the IRR/NPV distribution: “that’s the truth. Nobody can say, ’now that I see it, maybe we were pessimistic.’ Too late.”
• On edge: “We don’t win because we’re good at predictions. We win because we can run them better, so we can offer a good price.” Twelve years running businesses “from the trenches” beats PE’s weekly management calls — “you think you understand; I don’t think you really do.”
• Negotiation is Buffett-style: a near-maximum fair offer immediately, no haggling reputation. They bid on roughly twice as many companies as they buy and “have never lost a bid” — the misses were sellers who didn’t sell at all — which he reads against himself: “tells me we’re probably not very good at negotiating. Some level of failure rate would indicate a more optimal strategy.”

7. Evernote and AOL: “tarnished” brands with a next level to unlock

• Evernote was the leap from asset deals to structured companies — Ferrari admits “hesitation and insecurity,” like moving from local tennis to an international tournament. He guessed they paid ~50% more than the next best offer (“any good strategy needs to be somewhat win-win”) for a slightly tarnished brand a quarter of a billion people had used. In 2.5 years: ~250 significant product improvements, innovating “three to five times faster” with a smaller team, the code base and cloud rebuilt so notes sync in under 10% — sometimes 1% — of the old time.
• Pricing: ~60% higher on average; ~10% of lukewarm customers left, but “retention is at an all-time high” and satisfaction best ever. And sophistication isn’t synonymous with raising prices — at Meetup they introduced a free organizer tier while charging dedicated users more: “better at segmentation… whether that translates into higher prices or lower prices, I don’t know.”
• AOL, superficially “legacy, old, probably worth nothing,” is actually “a wonderful business”: tens of millions of loyal users, which Ferrari believes is “the fifth most used email inbox in the Western world,” with a better P&L than larger, buzzier messaging companies that “don’t even begin to compare.”
• Echoing Patrick’s mentor’s “12 things where there should be three”: “people in general overestimate the value of R&D… a very small number of things pay off handsomely and most things are a waste of money.” Evernote now has a lower cost base, yet nine of ten power users would call it better — because they built what customers “painfully needed.”

8. Scar tissue: the viral peak and the Grindr all-in

• One acquisition closed exactly at the peak of a viral wave; assumptions “completely changed” and returns were “drastically inferior.” The lesson: be “absolutely paranoid” about user-acquisition sources — either the value sits in already-acquired users, or the acquisition drivers must be predictable. Word-of-mouth under normal conditions: predictable. Viral moments and future paid-acquisition rates: not confidently predictable.
• Another lesson: Ferrari says they went after what was likely Grindr, the LGBTQ+ dating app, in 2019. CFIUS was forcing the Chinese owner to sell, and the deal would have quadrupled the company. Nine months of Ferrari’s time — with an M&A team of one — ended with a rival offering “a bit more” while Bending Spoons had “capped out on available sources of funds.” The real cost was concentration: “we put all we had into making that one thing happen. It didn’t happen. We hadn’t done anything else” — and that showed in those couple years’ growth. Now they’re “almost obsessive about seeing the world in terms of statistics.”

9. Financing and investors: lenders are visionary, entrepreneurs are entitled

• The capital stack: five years of pure reinvested earnings, then commercial bank debt at no more than ~3.5x trailing EBITDA “on a good day.” Equity raises mostly funded employee secondaries (four or five since 2019, every 12-24 months); total dilution ~10%. The latest: $700M at an $11B valuation, the largest round of any private company in Italy’s history. “If you really believe in what you’re doing, it should be painful to increase your capital base.”
• On the big debt raise: a lender’s upside is a 3-5% spread, so “it’s all about not losing it” — but that paranoia “breeds a thoroughness, a thoughtfulness that’s not always the case with equity investors.” His surprise: “a lot of lenders are actually quite visionary… I thoroughly enjoyed my conversations with lenders at least as much.” Debt markets are gigantic, standardized, and “vastly more efficient” than VC.
• He prefers permanent capital (“they don’t have to ask you to liquidate — incentives can become a little bit perverse”) but sides with investors against founder entitlement: acting shocked that institutional money wants a return within a timeframe “is either naive or intellectually dishonest.”
• His taxonomy of investors: the good ones are pattern recognizers; “the bad investors and the amazing investors — none of which is a pattern recognizer.” The truly outstanding grasp “almost the laws of physics” of a business and find what’s good that few think is good — anything pattern-fitting is already priced (“a lot of companies today in AI… most, even the good ones, are probably too expensive”). And judging people is the choke point: someone eight-out-of-ten smart “can only discern the sevens from the sixes… the nines, the 10s for them look like the same” — the early Amazon investor he knows bet primarily on Bezos’s “clarity of thought.”

10. AI, moats, and why nobody copies this — plus the kindest thing

• AI in the medium term is “mostly a good thing” for Bending Spoons: most units are 20% of revenue or less, so even one or two declining fast is “undesirable, but not existential” against 75% annual growth. AI accelerates quality and efficiency “if used properly, but it doesn’t do it by itself” — it takes custom integrations, proprietary tech, cultural work — and the gap between leaders and laggards “will widen for years.”
• On SaaS disruption, hedged precisely: telling ChatGPT “build me Jira” is “not months away, it’s not even a couple of years away.” Something that works the same 95% of the time is “infinitely easier” than something that works essentially 100% of the time; it’s not easy for users to specify what they need; and software is a small share of wallet — “a lot of stars need to align. It’s probably many years out.”
• Why no second Bending Spoons? PE’s barriers are low — “you just need to do well enough that you don’t look bad” — hence a “proliferation of wannabes.” But rebuilding this platform, even with $1B and full knowledge, would take him “easily seven or eight” years of “cultivating the little garden.” “There is no shortcut and I’ve seen nobody try cuz they just understand it’s just too painful” — which is exactly why he’s comfortable being this transparent.
• The closing story, worth the whole episode: pathologically shy as a child — “let’s say diagnosed with autism” — he spoke to nobody through his first year of middle school, until two popular classmates hugged him on a school trip and invested in him for months. Years later one revealed, in anger, that a teacher had asked them to help. “I’d never felt more grateful in my life… they literally changed my life.”