CS Holding — Weekly Focus · August 14, 2026

Nvidia persuaded six of the world's largest alternative managers that a GPU is collateral, and mobilised five hundred billion dollars without putting one of them on its own balance sheet. When money finds new security, the edge moves somewhere else.

In the same week Lovable doubled to a 13.3 billion dollar valuation in seven months, pushing the cost of internal software toward nothing. In Italy, Osservatorio PEM counted 318 deals year to date, up 36 percent, with add-ons making up 55 percent of buyouts. And a searcher closed at three times earnings that had already grown fifty percent since the letter of intent.

01 WILDCARD USD 500B

Nvidia mobilises 500 billion with six managers and pledges GPUs as collateral

Six of the largest alternative asset managers signed memoranda with Nvidia to mobilise more than five hundred billion dollars of third-party capital into AI data centres, with GPUs pledged as security. Nvidia may support up to twenty-five percent of residual value on any individual financing, project by project, keeping the buildout off its own books.

This is equipment leasing logic applied to compute: lend against an asset you assume can be resold. Anyone who has financed machinery knows where that formula breaks, and it is residual value. Read it the other way round as well. When the supplier guarantees the buyer's collateral, demand stops being entirely external.

02 AI USD 13.3B

Lovable doubles to 13.3 billion in seven months and internal software stops being capex

The Swedish company raised four hundred million dollars at a 13.3 billion valuation, double its 6.6 billion mark in December, with Menlo Ventures and EQT's Scaleup Europe Fund leading the round. Sixty million projects sit on the platform, and the apps built there draw nine hundred million visits a month.

For anyone running an SME the valuation is not the point. The line item is. The internal system, the configurator, the supplier portal: these used to be quarters and capex, now they are afternoons. Google halved Gemini Flash pricing in the same week. The quote you turned down in March deserves a rerun, not a file.

03 M&A ITALY 55% ADD-ON

Italian private equity runs at plus 36 percent, and half of all buyouts are add-ons

Osservatorio PEM's July read, carried by Milano Finanza, puts 318 transactions year to date against the same period in 2025, up 36 percent. Of July's deals, eighty-five percent were buyouts, and fifty-five percent of those were add-ons. Foreign investors signed 55 percent of the month's activity, Lombardy absorbed more than a third and Lazio close to twenty percent.

What matters for a searcher is who sits on the other side of the table. If more than half of Italian buyouts are additions to a platform that already exists, the competition on a family business is not another searcher. It is a fund with a consolidation mandate, priced synergies and a chief executive who signs them. You compete on speed and on the relationship with the seller, not on price.

04 SEARCH FUND 3x TO 2x

He paid three times earnings for a business that was worth two by closing

Boerner bought a twenty-eight-year-old Californian business that restores church pews for roughly 1.15 million dollars, three times owner earnings, funded with an SBA loan and fifteen percent down, no seller note. Between the letter of intent and closing, those owner earnings moved from four hundred to six hundred thousand dollars, taking the effective multiple to about two.

The multiple was not set by the business. It was set by how far behind the seller's own view of it had fallen. That is the same asymmetry sitting inside every Italian company with no successor and no adviser running a process. Whoever reads the current numbers first buys at a price that no longer exists. Year one closed near two million in revenue, headcount from seven to thirteen.

BIG PICTURE

Capital found new security in compute, and custom software is now close to free. Yet the two trades that actually made money this week turned on information: what a business is worth today, and who knows it first. In a market where half of all buyouts are planned consolidation, the edge is not capital. It is the current number.

On the last deal you looked at, how stale were the seller's numbers against yours? And how often did you notice before they did?