In January, Perplexity signed a three-year $750 million cloud agreement with Microsoft. Bloomberg reported it on January 29.

Divide by three. That is $250 million a year committed to one cloud vendor, at a moment when the whole company was running under $250 million annualized.

Eight months later, revenue has tripled and Nvidia is reportedly in talks to invest at more than $30 billion.

Committing capacity before the demand exists

I want to give that January decision its due before getting to the valuation, because it is the hardest call an AI infrastructure business makes.

Reserve compute early and you are on the hook whether or not the customers arrive. Wait until demand is proven and the capacity is spoken for, so you cannot serve the demand you just proved. There is no version of this where you get to be careful.

Perplexity took the first option with a number equal to its entire yearly revenue. The cargo showed up.

Worth holding onto that when the valuation headline gets passed around, because the headline makes it look like a story about hype and the underlying operating decision is more interesting than that.

What is reported versus what is settled

The Information published on Sunday that Nvidia is discussing an investment in a multibillion-dollar round that would value Perplexity above $30 billion. The prior round, about a year ago, put it near $20 billion.

Perplexity declined to comment. Nvidia did not respond to a request for comment.

So this is a reported conversation and not a signed deal. That label deserves to stay attached, because within two weeks the figure will circulate as settled fact, the same way several other rounds did this year.

The structure is verifiable regardless. Nvidia sells silicon to the industry and is weighing an equity position in one of the buyers of that silicon. We looked at the much larger version of that arrangement in Nvidia’s $105 billion guarantee on OpenAI leases. Smaller money here, same shape.

At roughly 40 times revenue, the valuation also carries an obligation. Companies priced that way have to grow into it, and the levers available are well known.

Why this matters to a company that invests in none of it

If you sell B2B, Perplexity stopped being a tool for technical people. It is one of the places your buyer finds out who does what you do and decides who to email.

A channel priced at 40 times revenue has to monetize, and the menu is short. Advertising. Harder limits on the free tier. Paid publisher agreements. Changes to how sources get selected and shown.

That last one lands on you directly. Right now your content can be cited in an answer with no commercial relationship of any kind. Nothing guarantees that survives contact with a $30 billion valuation that needs justifying.

Search ran this exact play between 2005 and 2015, and plenty of marketing teams lived through it. Organic traffic was free. Then free below three ads. Then free below a panel that already answered the question.

On what goes wrong when you trust how a model cites you, we wrote up the PwC hallucinated citations case. This is the other half of the same channel. It can cite you wrong today and change who it cites tomorrow.

What to actually do on a normal Monday

Two small things.

Measure what answer engines already send you, even if the volume is trivial and the measurement is rough. If it is 2% of sessions, you want to have known it was 2% before it changes. Without a baseline you cannot tell an algorithm shift from a slow quarter.

And do not build anything load-bearing on a single channel you do not own. True for Perplexity, true for Google, true for Facebook in 2014, true for LinkedIn right now. Borrowed reach is always borrowed.

Perplexity bet its whole revenue that demand would arrive, and it did. Your equivalent bet is smaller and much cheaper to hedge. Start by finding out how much of your discovery already runs through there.

Let us measure how much of your discovery runs through answer engines