Broadcom Doesn't Buy Companies. It Buys Customers Who Can't Leave.

Within weeks of closing its purchase of VMware in late 2023, Broadcom did something that looked like vandalism to anyone running a data center: it killed the perpetual license. The thing IT departments had bought outright for two decades - software you owned, like a tool in a drawer - was gone. vSphere, vSAN, NSX: rent only, billed by subscription, take it or migrate.⁸ Some customers opened their renewal quotes and found the price had gone up as much as fifteen-fold.¹¹ This was not a glitch in the integration. This was the integration.

The official story is that Broadcom is a diversified technology company, building scale across semiconductors and enterprise software. That framing misses the engine entirely. Broadcom does not buy companies to grow into them. It buys companies whose customers cannot leave - and then tests, methodically, exactly how much those customers will pay before they finally do.

The target is captivity, not technology

Look at the shopping list and a pattern resolves. CA Technologies, the mainframe-software backbone of large enterprises, for $18.9 billion in 2018. Symantec's enterprise security business for $10.7 billion in 2019.⁴ Then VMware. None of these are growth assets. They are installed assets - software so woven into how a bank or an airline runs that ripping it out is a multi-year project with its own risk of catastrophe. That is the whole point. Broadcom is not buying revenue; it is buying switching costs. The more painful it is to leave, the more pricing power the buyer inherits on day one, before a single line of code is improved.

Once you own a captive base, the mechanism is almost mechanical. Step one: convert ownership to rent. A perpetual license is a relationship you can walk away from; a subscription is a tollbooth you pass through every year. Step two: bundle. Force customers buying one product to license a suite they don't fully use, so the metered price rises even where consumption doesn't. Step three: raise the price - and keep raising it - until the churn it triggers finally outweighs the revenue it captures. The genius and the danger live in the same move: Broadcom prices not to the value it delivers, but to the cost of leaving.

What the $61 billion headline gets wrong

Almost every retelling says Broadcom paid $61 billion for VMware. That number is a fossil - it's the equity value on announcement day in May 2022.² By the time the deal actually closed on November 22, 2023, Broadcom's own stock had appreciated sharply, and the 544 million Broadcom shares handed to VMware stockholders were worth about $53.4 billion at close, on top of $30.8 billion in cash. The real transaction value, recorded in Broadcom's SEC filings, was roughly $84.2 billion.¹ That $23 billion gap matters, because it explains the urgency of the extraction. When you pay $84 billion and book roughly $54 billion of it as goodwill¹⁰ - accounting language for 'we paid far more than the assets are worth on paper' - you do not have the luxury of nurturing the customer base. You have a clock running, and the price hikes are the clock.

The bill the model hides until it's due

Here is what an extraction engine doesn't show you on the income statement: the customers quietly deciding to leave. In the first half of 2024, Gartner inquiries about VMware alternatives spiked 275% year over year.⁷ Roughly half of VMware's midmarket base had already let licenses lapse around the acquisition.⁷ These are not complaints; they are the early frames of a migration. Squeeze a captive customer hard enough and you don't just raise this year's revenue - you fund their escape budget and hand them the motivation to spend it. The free cash flow looks magnificent right up until the renewal that doesn't come. Broadcom's roll-up records its wins instantly and its losses on a multi-year delay, which is exactly why it's so dangerous to admire from the outside.

Isn't this just disciplined capital allocation?

The fair objection is that Broadcom is not a villain; it is a clear-eyed allocator of capital, and the market has rewarded it for years. Two honest points cut against the cynical read. First, the subscription shift wasn't even Broadcom's idea - VMware itself began the transition in 2018, as Hock Tan was quick to point out while defending the changes.⁸ Broadcom accelerated a direction the industry was already moving. Second, plenty of customers will pay the higher price, because the software genuinely is mission-critical and the alternative genuinely is worse; extracting value from an asset you overpaid for is not a scandal, it's the job. All true. But the steelman concedes the real point. The strategy only works if the cost of leaving stays higher than the price of staying - and every quote multiplying costs eightfold or more¹¹ narrows that gap by motivating the very migrations that destroy it. Discipline that erodes its own moat isn't discipline. It's harvesting the orchard for firewood.

Broadcom has been disciplined about one thing above all: knowing what it is buying. Not technology, not growth, not even a brand - it buys the distance between a customer and the door. The roll-up runs beautifully as long as that distance stays long. But a regulator already drew Broadcom a hard line once, blocking its $117 billion run at Qualcomm on national-security grounds in 2018, one of only five such presidential vetoes in thirty years.⁵⁹ The VMware backlash is a different kind of veto - the kind customers cast with their migration plans, slowly, and then all at once. The genius of the model was never the pricing. It was choosing assets nobody could leave. The risk is that 'nobody can leave' has an expiration date, and Broadcom keeps testing exactly where it is.