Bending Spoons buys software companies, integrates them on one platform, and runs them on a ten-year horizon to fund bigger deals.
Ferrari says AI wrote 94% of Bending Spoons' code, up from zero two years earlier, and no good company should be below 70%.
Bending Spoons keeps leverage below 3x, refinances debt yearly, and hedges 100% against rate moves.
Bending Spoons buys software companies and compounds their cash
Bending Spoons is a compounding machine built on acquisition rather than invention. It buys software businesses it thinks it can run far better, moves them onto one platform and one core team, and uses the resulting cash flow, levered below three times, to buy bigger ones. Ferrari's bet is that established software keeps its moats even as writing code gets cheap, and that a decade-long horizon beats chasing radical reinvention.
Takeaways
Bending Spoons compounds cash flow through acquisitions. Buys technology companies it can make far more successful, integrates them onto one platform with the same foundational technology and core team, and runs them on a ten-year view. The extra cash flow is levered into bigger targets. Roughly 50 acquisitions, maybe 60.
Debt works when default risk is negligible. Ferrari keeps leverage well below three times, currently 2.4 to 2.5 and historically closer to two. He refinances most debt every year so maturities stay five or more years out, hedges 100% of it against rate moves, and cites Broadcom and TransDigm.
Radical invention has a terrible median outcome. Ferrari calls the bond versus option framing simplistic but directionally true. Radical bets offer higher peaks and a median where you lose everything. Incremental innovation is dependable and measurable, which makes Google's reliance on search defensible, though Acharya notes Google also produced the transformer paper.
Cheaper software did not break established moats. Replication was already cheap. Products endure on first-mover advantage, brands, habits, network effects and switching costs, such as Eventbrite's 90m monthly ticket-seekers or Komoot's proprietary trail data. Most Airtable customers pay under $20,000 a year, and 0.2 FTE of an engineer would cost more.
Managers avoid risk because the median hurts. Ferrari says executive teams shun risk because they hate the median case, and draws the venture analogy of four or five companies in 50 returning 100x. Diversification depends on which phenomenon you are hedged against, since supposedly separate assets can be highly correlated.
Aim high and hire talent-dense teams. Ferrari's closing advice is ambition, rationality, intellectual independence and talent density, since density is nearly impossible to improve if you start low. He kept his failed pre-Bending Spoons startup Evertail alive for three years because he is poor at judging when to quit.
The operating system inside Bending Spoons
More than 50 tools built in-house, covering data processing, A/B testing, marketing automation, user lifetime prediction, credentials management and recruiting.
Tap a name to go there · tap a line to see why they are linked
On this episode 3
Show 1
ALT, an agent in Slack, automatically holds the same tool and data access as the employee, runs analysis, and escalates to IT security or credentials teams.
Customer support is essentially all AI or external vendors, and at least 90% of data fetching and cleaning is done by AI.
AI wrote 94% of the company's code at the last check, up from zero two years earlier. Ferrari says no good company should be below 70%.
Ferrari says the tools only pay off with the right culture and talent, since the user makes much of the difference.