Broadcom Didn't Buy VMware to Grow It. It Bought a Captive Audience.

An AT&T executive sat down to renew the software that runs much of the company's virtualized infrastructure and got a number back that read less like a quote and more like an eviction notice: a 1,050% increase, written into an email that later landed in a court filing.⁴ AT&T sued. It also settled quietly and on undisclosed terms.⁵ That sequence — outrage, lawsuit, settlement, stay — is the entire story of what Broadcom did to VMware, compressed into one customer who happened to be big enough to fight back.

The official story is that Broadcom bought VMware to build the next great enterprise-software platform. The real story is that Broadcom bought a base of customers who can't easily leave, and is now collecting on the fact that they can't.

Broadcom didn't acquire VMware to make it better. It acquired VMware to make it pay — and the playbook for doing that was written long before the deal closed. Strip out perpetual licenses. Bundle everything into subscriptions. Cut the cost base in half. Then point all of it at the one variable that matters: how much you can charge people who have nowhere to go.

The price was never the asset. The lock-in was.

Start with what Broadcom actually paid for. The deal valued VMware at roughly $61 billion in equity, with another $8 billion of net debt assumed on top — call it $69 billion all in.¹ That is a staggering price for a company whose growth had stalled. It only makes sense if you stop thinking of VMware as a growth asset and start thinking of it as an annuity that has been chronically underpriced. Inside that figure sits the real target: in the same May 2022 announcement, Broadcom told investors it expected to add about $8.5 billion in pro forma EBITDA from VMware within three years of closing.³ That number is the thesis. Everything Broadcom has done since is in service of hitting it.

Here is why the math works. VMware's virtualization software sits underneath the data centers of most large enterprises on earth. Ripping it out is not a procurement decision; it is a multi-year migration involving re-architecting workloads, retraining staff, and risking downtime on systems that cannot go down. The switching cost is the moat — and the cruel arithmetic of a moat is that the harder it is to leave, the more you can be charged for staying. Broadcom didn't invent that captivity. It simply recognized that VMware's previous owners had been far too gentle in monetizing it.

Kill the option to stay still

The first move was to take away the exit that wasn't an exit. For years, VMware sold perpetual licenses — buy once, own forever, pay only for support if you wanted updates. That model gave customers a quiet form of leverage: at renewal time, they could always shrug and keep running what they already owned. Shortly after closing in late 2023, Broadcom eliminated perpetual licenses entirely and moved all new sales to subscription-only terms.⁶ Overnight, 'I'll just keep what I have' stopped being an answer. The clock now runs on Broadcom's terms, and when the term ends, you renew at the new price or you migrate — and migrating, as established, is the thing you cannot easily do.

It's worth being precise here, because the subscription shift is often misread. VMware had already begun moving toward subscription licensing before Broadcom arrived; Broadcom didn't originate the transition. What it did was accelerate it, mandate it, and weaponize it — pairing the licensing change with aggressive bundling that pushes customers into larger, costlier packages than the discrete products they used to buy. The subscription model wasn't the cruelty. It was the delivery mechanism for the cruelty.

Cut the cost base in the shadows

Raising the top line is only half the EBITDA equation. The other half is the cost base, and here Broadcom moved with a discipline that bordered on the surgical. Reporting indicates that in the roughly two years after the acquisition, Broadcom eliminated something like 19,000 VMware jobs — close to half of the company's pre-acquisition headcount of about 38,000.⁷ What makes that number instructive is not its size but its choreography. WARN Act filings, the legally required notices for large layoffs, captured only around 2,000 of those cuts.⁷ The rest came in rolling monthly batches — a slow drip rather than a single announced reduction — leaving no single headline to read 'half of VMware gone.'

Put the two halves together and the machine is obvious. On the earnings call after closing, Hock Tan projected VMware would contribute about $12 billion in revenue in fiscal 2024.⁸ Multiply locked-in customers by far higher prices, subtract a workforce cut roughly in half, and you arrive — quickly — at the $8.5 billion EBITDA target. This is not a turnaround. It is a harvest.

The fair objection: maybe VMware was just badly run

The honest counter is that VMware really had been undermanaged, and Broadcom is not wrong to fix it. Mature enterprise software that has stopped growing genuinely is overstaffed relative to what it ships; perpetual licenses really do leave money on the table; and the subscription transition was already underway before Broadcom touched it. By this reading, Broadcom is simply a ruthless but rational operator pricing a critical product at what it's actually worth and cutting the slack a sleepier owner tolerated. There is truth in that. A vital piece of infrastructure being sold too cheaply is a real inefficiency, and someone was always going to correct it.

But notice the tell. The same playbook — strip licenses, bundle, slash staff, raise prices on the trapped — is the one Broadcom ran at CA Technologies and at Symantec before it.¹⁰¹¹ When a strategy is identical across three acquisitions, it is not a diagnosis tailored to one company's problems. It is a template applied to whatever it buys. And a template optimized for near-term EBITDA extraction trades away the thing a software franchise actually runs on: the engineers who build the next version, and the customers who trust you enough to stay without a lawsuit. Broadcom is collecting on goodwill it is also spending. The question is only which runs out first — the lock-in, or the patience of the locked-in.

Broadcom paid $69 billion not for software but for an audience that cannot walk out of the theater, and it is now charging admission accordingly. The migrations are real but slow; the lawsuits are real but settle; the price increases are real and they stick — for now. That is the bet: that switching costs decay slower than customer loyalty does, and that by the time the captive base has engineered its escape, the EBITDA will already have been booked and the next acquisition already begun. Broadcom isn't building the future of VMware. It's monetizing the past, on the clock, before the room empties out.