Three gatekeepers went at once
Distribution used to be a problem with known solutions. You bought ads, you hired an agency to place stories, or you ranked. Each route had a gatekeeper, each gatekeeper had a price, and the price was legible.
All three degraded inside roughly the same window, for unrelated reasons.
- Paid stopped reaching the buyer. The technical audience that evaluates AI tooling pays monthly for the version of the feed without promoted posts. You can still buy impressions. You cannot buy the specific person deciding what their team evaluates, because they have opted out of the inventory. The mechanics.
- Trade press stopped being read by the buyer. The placement still happens and the audience that used to justify it has dispersed into feeds, newsletters and group chats. A story in a publication your buyer does not open is a credibility artifact rather than a distribution event, which is a real thing to want and a different thing from what you paid for. The longer version.
- Search started answering instead of listing. A buyer asking which tools handle their case now reads six sentences naming three companies and never sees a results page. Ranking first matters less when the click does not happen. What replaces it.
What survived is the thing none of the three could be reduced to: a person your buyer already chose to follow, saying something. That route has no auction, no rate card and no gatekeeper you can pay to skip, which is exactly why it is the one that still works.
The signal from the capital side
The most useful evidence for any of this comes from people with no incentive to argue it.
Andreessen Horowitz now runs what is functionally a media company inside a venture firm. By their own published account: roughly a million followers on X, a daily newsletter at 250,000 subscribers, around a million podcast downloads a month, and a growing Instagram presence.
Commentators tracking the effort put the marketing headcount above 70, against the one or two people most firms carry. They produce launch films, essays and launch pushes for portfolio companies, and they describe the whole thing as going direct, offered as a service.
Read past the specifics and there is one claim underneath it: a firm whose product is money decided that money was not the scarce part. When capital is abundant and reach is not, the firm that can guarantee a founder will be heard is offering something the firm writing an identical cheque cannot.
The same shift shows up further down the market without the budget. Founders who post are raising on audiences rather than decks.
Categories are being defined by whoever explained them publicly first. None of that required anyone's permission, which is the part that makes it durable.
The sequence inverted
The standard order was capital, then company, then market. You raised, you built, and you went to find the people who wanted it.
For a growing set of companies the order now runs the other way. Someone builds an audience around a problem, learns precisely what that audience wants, builds it, and takes capital afterwards against a market they can already demonstrate.
The audience is the asset that makes the rest cheaper: cheaper to validate, cheaper to launch, cheaper to hire into, cheaper to raise against.
Two consequences worth sitting with.
- A seed deck without a distribution answer now looks incomplete. Not a marketing slide listing channels. An answer to how the first ten thousand people find out, and why that will keep working.
- Distribution moved from a cost line to an asset line. An audience, a set of relationships with people who can carry a message, and a description of you circulating on other people's sites are all things that keep producing after the spend stops. The compounding argument.
Neither order is universal. Plenty of excellent companies still raise, build and sell in the old sequence, and deep infrastructure often has to. The claim is narrower: the second sequence stopped being unusual, and any strategy assuming it is unusual is now working from a stale map.
Two ways to hold distribution
Once you accept that reach is the constraint, there are exactly two ways to have it, and most of the confusion in this subject comes from treating them as one thing.
| Own it | Borrow it | |
|---|---|---|
| What it is | Your accounts, your list, your podcast, your search positions | Reach through people who already hold an audience |
| Cost shape | Time and headcount, paid continuously, returns arriving late | Per campaign, immediate, stops when you stop |
| Ceiling | Credibility. Everyone knows who pays for your channel. | Availability. You get whoever is free and right. |
| Timeline | Years at the scale that matters | Days |
The reason the big media build is interesting rather than intimidating is that it is one firm's answer to this table, chosen because they could afford the expensive column. Most companies cannot, and the useful question is not how to match it. The decision in full.
What it does to the org chart
The structural change nobody budgets for: distribution needs an owner, and at most companies it does not have one.
At ten people the founder owns it by default. At two hundred there is a function with a budget.
In between there is a product marketer, a demand-gen lead and a content hire, each with a sensible scope, and a hole between them where the question of whether anybody sees the thing is nobody's first priority. Who owns it.
The one part that does not delegate is the founder's own account. Every version of this that works has a person standing behind the claim, and a brand account posting a strong opinion reads as a committee that approved a strong opinion. Why that outperforms.
Where it goes next
The change already in progress is that distribution has started compounding into machine-readable reputation.
When a buyer asks an answer engine which tools do the thing, the system runs a live search, pulls a set of pages, and writes three names into a sentence. Those pages are not yours.
They are the newsletters, forum threads, breakdowns and posts written by other people, and the description they settled on becomes the description the model repeats. The mechanics are here.
So the value of a launch now has a second half that arrives months later. The reach is the immediate return. The residue is a set of independent pages describing your company in roughly the same words, which is the input to every automated answer a buyer receives from that point on.
A company with no distribution is now absent from the machine layer that buyers ask first, as well as unknown to the buyers themselves.
The second-order effect is that consistency of description becomes an infrastructure decision. Eleven people describing you eleven ways produces a vague answer or somebody else's category. The sentence is worth settling before the launch, because a launch is the single moment when the largest number of people write it down at once.
What has not changed
Three things, and each of them is the reason the rest is worth doing carefully.
The product still has to be good. Distribution accelerates the discovery that something is not ready, and at any real scale the people who make that discovery have long memories and public accounts. This is a genuine risk and worth saying out loud in the planning meeting rather than after.
The claim still has to be arguable. Reach applied to a statement nobody could disagree with produces impressions and nothing else. The floor is creators multiplied by their median reach; everything above the floor is decided by whether the thing being carried was worth carrying.
Somebody still has to be accountable for it. The whole mechanism runs on a reader believing a person meant what they said. That does not survive being automated, outsourced or approved by committee, which is the constraint that keeps this from becoming another channel with a rate card.
Questions we get asked
Why are venture firms building media operations?
Because capital stopped being the thing that is hard to get and attention started. When money is abundant and reach is scarce, the firm that can guarantee a founder will be heard is offering something the firm writing a similar cheque cannot. Andreessen Horowitz reports around a million followers on X, 250,000 daily newsletter subscribers and roughly a million monthly podcast downloads, which is a media company sitting inside a venture firm.
Has the go-to-market sequence actually changed?
The order has inverted for a growing set of companies. Raise, build, then find a market was the standard sequence. Increasingly it runs the other way: build an audience, learn what it wants, build that, and let the capital follow a demonstrated market.
Neither order is universal, and the second one is now common enough that a seed deck without a distribution answer looks incomplete.
Does this mean paid advertising is finished?
For technical B2B buyers it was structurally weak before any of this. A large share of them pay monthly to remove promoted posts from the feed where they spend their attention. Paid still works for consumer categories and for retargeting people who already know you, and it was never the route to an engineering leader deciding what to evaluate.