What a $250k billboard says about scarcity

On 3 August the San Francisco Standard published the going rate for the billboards on the US-101 “skyway” between SFO and the Bay Bridge: $200,000 to $250,000 a month for the most prominent sites — about $3 million a year for one face. Mid-tier skyway positions rent for $50,000–$100,000 a month. Prices are 20–30% above pre-pandemic levels, and nearly every premium site is leased by an AI, tech or finance company.

It is a fun story about an advertising arms race. It is also the clearest available demonstration of how out-of-home pricing actually works — and the mechanism has nothing to do with impressions.

What happened

Top-tier US-101 billboard sites in San Francisco now command $200k–$250k per month, roughly $3m a year for a single face (SF Standard, 3 Aug 2026 — trade reporting, sourced to named brokers).
Mid-tier skyway positions: $50k–$100k per month (SF Standard — directional).
Premium prices are up 20–30% versus pre-pandemic, per Eric Perko, founder and CEO of Apollo Partners (named source — directional).
Supply is fixed and effectively locked: a short stretch of highway, with companies holding sites on multi-year leases (SF Standard — directional).
Demand is concentrated in AI, tech and finance — WorkOS, Nebius and Lambda among the named advertisers — targeting the 100,000+ daily commuters on the corridor (SF Standard — directional).
Perko’s own caveat: “It certainly can’t last forever. There are cycles for everything.” (named source).
The read: the price is set by non-substitutability, not by audience volume — and a beauty network’s most valuable asset is also a position, not a headcount (BDOOH analysis).

What’s actually being priced

A $250,000-a-month billboard does not deliver forty times the audience of a $6,000-a-month one. It delivers the only place where a specific audience, in a specific state of mind, can be reached at all. Three properties are doing the pricing work:

Fixed supply. The skyway is a few miles long and every good face on it is spoken for, often for years. No amount of demand creates a new one. When supply cannot respond, price absorbs the entire increase.

Non-substitutability. A buyer who wants “the sign every venture capitalist and engineer in the Bay Area drives past” has no alternative purchase. Cheaper inventory elsewhere is not a discount on the same thing; it is a different thing.

A homogeneous, urgent buyer pool. AI companies are competing with each other for the same recruits, the same customers and the same narrative. That’s a bidding war by construction — and Perko’s “it can’t last forever” is the tell. A rate held up by one hot category is a rate with a cycle attached to it.

Notice what isn’t in the pricing logic: CPM. Nobody is buying $3m of impressions. They are buying a position, and the impressions are a by-product used to justify it after the fact.

What it means for beauty

Beauty networks price the wrong noun. Almost every media kit we see prices screens, per screen, per month, as if the estate were interchangeable stock. But the salon mirror in front of a client mid-service is a non-substitutable position in exactly the way a skyway billboard is — there is no other way to reach that person, in that state, at that moment. That’s the argument for a tiered rate card built on position and daypart in how to price your inventory and packaging and pricing for advertisers.
Scarcity is the only pricing power a small network has. A hundred salon screens will never win on reach; the cross-format reach and frequency maths is unkind to small estates. What a small estate can own is exclusivity — category exclusivity, venue exclusivity, moment exclusivity — which is why running a founding advertiser program sells better than a rate card ever will.
A single-category demand base is a risk, not a milestone. San Francisco’s premium inventory is priced by AI money and will reprice when that cycle turns. A beauty network funded entirely by endemic beauty brands has the same fragility in miniature — the concentration problem we work through in endemic vs non-endemic, decided and the endemic advertiser map, and one of the recurring causes in why beauty DOOH networks fail.
“Advertising in real life is booming” is the tailwind, and it is real. Perko attributes the demand to brand-building and to distrust of digital advertising — the same structural case we make in why OOH amplifies digital and brand safety: physical vs digital. Beauty venues sit squarely inside that argument; they just have to be buyable when a planner acts on it.

The caveat that keeps us honest

This is trade reporting, not a rate card: figures come from named brokers and an operator sales director describing a market they participate in, and posted asking rates in premium out-of-home are routinely discounted in practice. The prices are specific to one corridor in one city during one category’s boom, and they say nothing about what any other inventory is worth. No beauty figure appears here and none is implied — there is still no reliable public beauty CPM, and the programmatic OOH CPM tracker records the only transacted prices we consider comparable. What transfers from San Francisco is the mechanism, not the money.

Sources

The San Francisco Standard — “One billboard, $3M a year: Welcome to AI’s advertising arms race,” 3 Aug 2026: top US-101 skyway sites at $200k–$250k per month (~$3m per year); mid-tier sites $50k–$100k per month; premium prices up 20–30% versus pre-pandemic; sources Eric Perko (founder/CEO, Apollo Partners), Michael Parvin (sales director, Outfront) and billboard tracker Dan Berte; advertisers named include WorkOS, Nebius and Lambda; ~100,000 daily Bay Bridge commuters. Trade reporting with named sources — directional; asking rates, not verified transacted prices. Secondary coverage: invidis, “OoH: AI Boom Sends San Francisco Billboard Prices Soaring,” Aug 2026. No beauty-specific rate, CPM or benchmark is asserted; beauty pricing method is developed in the Guides and Research.