Fox just agreed to spend more than 80% of its own enterprise value to buy Roku.
Fox reaches roughly 125 million people each day. It owns television networks, sports rights, news programming, studios, and decades of intellectual property. Roku owns the operating system that appears when more than 100 million streaming households turn on their televisions.
The deal valued Roku at more than 40 times EBITDA. Fox trades near eight.
A company that decides what millions of people watch concluded that the layer helping them decide what to watch next was worth five times more.
That looks like a media story. It is not.
The same trade is appearing in asset management, software, advertising, and financial markets. Across each of them, value is migrating away from the product that requires commitment and toward the infrastructure that monetizes the next decision.
The unit is shrinking
Consider Wealthfront and Robinhood.
Wealthfront manages roughly a third as many assets as Robinhood but is worth a tiny fraction of its market capitalization. Both serve individual investors. Both hold customer assets. Both operate under broadly similar regulations.
But they monetize different behavior.
Wealthfront is designed to help customers make a decision once: deposit money, select a portfolio, and leave it alone for decades.
Robinhood makes money from what customers do next. They trade a stock, buy an option, move into crypto, earn interest on cash, borrow against a position, or subscribe for additional services. Wealthfront primarily monetizes the balance. Robinhood monetizes the balance and the activity surrounding it.
The market appears to value the second model far more highly.
The same distinction is emerging in software.
Traditional SaaS asks a company to predict its needs in advance: how many employees will require a seat, which features they will use, and whether they will still need the product twelve months from now. The customer commits by the year and consumes by the month.
AI increasingly reverses that model. Customers pay per token, query, image, inference, or completed task. They do not have to predict next year’s usage before making today’s purchase.
Public markets are punishing software sold by the annual seat while private markets are assigning extraordinary values to software metered by consumption.
Netflix offers a softer version of the same signal. Its business continues to grow, but investors are paying less for each dollar of future earnings than they once did. Netflix must repeatedly persuade a subscriber to remain committed to one monthly service and then to spend hours inside its library. Roku can make money as the viewer moves among Netflix, YouTube, Fox, Disney, and whatever comes next.
One owns a destination. The other owns the intersection.
These are not identical businesses, and valuation gaps always have several causes. But the recurring pattern is difficult to ignore: the market increasingly rewards companies that shorten the interval between the customer’s decisions.
Why now
The cost of changing your mind has collapsed.
Capital can move between assets, strategies, and platforms in seconds. Advertising inventory is auctioned one impression at a time. Software can be called through an API and metered by the token. Entertainment no longer arrives in a scheduled block; it is selected, abandoned, and replaced continuously.
When the available menu changes faster, the option to wait becomes more valuable.
A year-long commitment now has to compete not only with the alternatives available today, but also with the alternatives that might appear three months from now. The faster technology expands the menu, the more expensive it becomes to lock yourself into one item on it.
This does not mean commitment itself has become worthless. The opposite is closer to the truth.
The winning businesses still make enormous commitments. They build data centers, aggregate liquidity, subsidize devices, acquire customers, license content, and assemble publisher networks. They simply avoid asking the customer to make an equally large commitment in return.
They buy their inputs wholesale and sell their outputs atomically.
That is the structure worth looking for.
Who wins
Hyperliquid is one of the cleanest examples.
Billions of dollars are deposited on the platform, creating a durable base of liquidity. That base supports trading volume that can equal or exceed the deposited capital each day. Hyperliquid earns a small fee each time someone trades.
The infrastructure is persistent. The revenue event is instantaneous.
A user does not have to commit to owning an asset for a year or even to holding the same view for an hour. Hyperliquid gets paid when the user enters, exits, changes direction, takes leverage, or admits that the previous decision was wrong.
The business is not tied to any particular forecast about where prices go. It is tied to how frequently people reconsider.
AppLovin has the same structure in advertising.
Its technology connects advertisers with a large network of mobile applications and reruns the market for attention impression by impression. Advertisers do not need to commit to one game, one audience, or one campaign for years. They can buy an outcome, observe the result, and bid again.
The user’s disloyalty is not necessarily a problem. A player leaves one game and enters another; AppLovin has another opportunity to sell the next impression.
The company’s durable assets—its data, publisher relationships, software, and advertising infrastructure—sit underneath a stream of tiny, repeated transactions.
Roku is moving toward the same position in television. It does not need one streaming service to hold the viewer forever. It needs to remain between the viewer and the next choice.
This is the distinction:
A weaker business asks the customer to commit for a long period and then hopes the customer uses the product.
A stronger one commits the infrastructure itself and gets paid whenever the customer acts.
The best businesses may therefore be neither traditional subscription companies nor purely transactional marketplaces. They combine a committed base with an uncommitted customer:
Durable infrastructure, variable consumption.
Persistent liquidity, momentary trades.
A permanent installed base, constantly changing attention.
Large fixed commitments, tiny units of revenue.
Nobody wants to commit. The best businesses commit on everyone else’s behalf—and charge them every time they change their mind.
What would prove this wrong
There are obvious alternative explanations.
Robinhood may simply be growing faster than Wealthfront. AI may be disrupting traditional SaaS rather than revealing a broader preference for consumption-based pricing. Roku may be worth more because of its first-party data and advertising inventory, not because it occupies a shorter decision interval.
The theory becomes meaningful only if the pattern survives after controlling for growth, margins, and competitive disruption.
Here is the test: businesses that monetize actions should continue to command a premium over businesses serving similar customers through seats, balances, or long contracts—even when their underlying growth rates converge.
If that premium disappears when interest rates fall or growth normalizes, then this was a cycle dressed up as a structural argument.
But if it persists, Fox’s decision will look less strange.
Fox was not buying a remote control.
It was buying the next decision.
This was good. Would point me to building an AI video platform in my field. Many things have been pointing me there recently