The Future of Advertising Papers

Paper No.26 · Attention Economics

Ambient Advertising: Marketing Inside Always On Computing

Pierre Subeh·June 26, 2026·8 min read

Abstract

Computing is dissolving into glasses, earbuds, cars, and rooms that never turn off. I map how advertising survives inside always on environments, where the old unit of the ad slot disappears and influence has to be earned as ambient presence.

The screen taught advertisers a comfortable lie: that attention arrives in rectangles. A rectangle has borders, and borders create slots, and slots can be sold. The entire ad economy, from the newspaper column to the story frame, is a business of rectangles. Always on computing erases the rectangle, and most of the industry has not begun to process what that means.

By always on computing I mean the environment that is already assembling around us in 2026: earbuds worn through the workday, glasses that annotate the street, cars that talk, assistants that listen in the kitchen, and an AI layer that persists across all of them. In this environment there is no session to open and no feed to scroll. The computer is simply present, the way electricity is present. You do not visit it. You live inside it.

Advertising inside a place you live obeys different physics than advertising inside a place you visit. Get it wrong and you are not an annoying banner, you are an intruder in someone's kitchen. Get it right and you achieve something screens never offered: presence at the exact moment of need, with zero competing rectangles. This paper is about the difference.

Key Findings

  • The ad slot, the foundational unit of media buying for a century, does not exist in ambient environments, and no amount of retrofitting will recreate it.
  • Ambient influence runs through three doors: being the default, being the recommendation, and being the sponsor of a genuinely useful moment.
  • Tolerance for interruption in ambient contexts is close to zero, which means the effective ad load of these environments will be a tiny fraction of the feed era's.
  • What I call the Presence Budget becomes the scarce resource: each brand gets a small number of moments per user per week before it is muted, and muting in ambient systems tends to be permanent.
  • Between 2026 and 2031, the brands that win ambient will be the ones that build utility worth summoning, not the ones that buy adjacency.

The End of the Slot

A slot requires three things: a bounded surface, a predictable moment of exposure, and a separation between content and commerce that the audience understands. Ambient computing destroys all three. There is no bounded surface when the interface is a voice in your ear. There is no predictable moment when the computer responds to your context instead of your schedule. And the separation between content and commerce collapses when a single assistant voice delivers both.

This is why I wince when I see agencies pitch "audio banners" for assistants or "spatial billboards" for glasses. These are rectangles smuggled into a world without walls. Some of them will ship, some will even get budget, and nearly all of them will train users to do one thing: find the setting that turns them off. In an always on environment, the off switch for advertising is discovered once and flipped forever.

The honest starting point is to admit that ambient environments will carry radically less advertising per hour than feeds did. The question is not how to cram the old load into the new space. The question is what the few surviving commercial moments look like and who controls them.

The Three Doors

Influence in ambient computing runs through three doors, and only three.

Door one is the default. When someone tells their car to order coffee, some brand fulfills that request without a menu ever being shown. Defaults in ambient systems are set by a tangle of platform deals, user history, and increasingly the judgment of an AI intermediary. The commercial battle for defaults will be the largest below the line spend category of the 2030s, and it will look more like enterprise sales and structured data work than like media buying. This is the same battle I described in marketing to AI agents: you win it with machine readable truth, verifiable reputation, and operational excellence, because the intermediary checks receipts.

Door two is the recommendation. When the user asks an open question, which option gets spoken aloud? An earbud answer has room for one name, maybe two. Screen search trained us to fight for ten blue links; ambient computing is a market where shelf space rounds down to one. Everything I have argued about answer engine optimization applies here with the volume turned up: citations, entities, and authority decide whether you exist at all.

Door three is sponsorship of useful moments. The assistant that offers a free navigation tip brought to you by a tire brand. The recovery summary after a run, presented by an electrolyte company. This door stays open only when the utility is real and the sponsor is lightly attached. The moment sponsorship distorts the utility, the user's trust in the whole channel dies, and the platform knows it, which is why platforms will police this door brutally.

The Presence Budget

Here is the framework I now use with clients planning for ambient channels. Every user grants every brand a Presence Budget: a small, personal allowance of moments in which that brand may appear in their ambient environment before irritation converts to muting. Unlike the feed era, where annoyance cost you nothing because exposure was forced, ambient systems give the user granular, durable controls. Spend the budget badly and you are not skipped. You are unsubscribed from someone's life.

Think of it as an illustration: suppose a typical user tolerates three unrequested brand appearances per week across their entire ambient stack before reaching for controls. Three. Against that allowance, an interruptive strategy is bankruptcy on a schedule. The only sustainable strategy is to make appearances the user would have requested anyway: the reorder reminder timed to actual depletion, the warranty note before the road trip, the price drop on the exact item they asked about. Ambient advertising converges on being a good butler, and a good butler is mostly silent.

This inverts creative priorities. The feed era rewarded attention seizure: pattern interrupts, hooks, loudness. The ambient era rewards judgment: knowing when not to speak. I expect the best ambient marketing teams of 2030 to look like CRM teams with editorial taste, not like content studios.

The Data Spine

None of the butler behavior is possible without a data spine the brand actually controls. Ambient moments are triggered by context: location, time, purchase history, stated preferences, device signals. Platforms will expose some of this context to advertisers, but always on their terms, always revocable, and always priced at extraction rates.

The alternative is consented, directly held data, which I have long argued is the only marketing asset that appreciates. A first party data strategy stops being a compliance hedge in the ambient era and becomes the literal fuel of the channel: you cannot time a useful moment for a person you know nothing about. My prediction, stated so it can be checked: by 2029, brands with rich consented data will pay materially less per ambient placement than cold buyers, because platforms will score advertisers on predicted user tolerance, and tolerance follows relevance.

What Breaks and When

Timelines, so this paper can be graded later. Through 2027, ambient remains a minority channel and most brands rightly treat it as an experiment. Between 2027 and 2029, glasses and earbud assistants reach mainstream commuter adoption in wealthy markets, and the first serious ambient budgets appear, mostly wasted on ported rectangle thinking. Around 2029 to 2031, the defaults market matures, presence scoring becomes an explicit platform mechanic, and the gap between butler brands and interrupter brands becomes visible in the data. If by 2031 ambient environments are carrying feed level ad loads and users are tolerating it, my whole model is wrong. I am confident they will not, because every prior interface transition has reduced tolerated ad load, never increased it.

What I Would Do About It

Start with the unglamorous work: entity data, structured product truth, reviews, availability feeds, and API accessible services. The three doors all check these credentials before they open. A brand that cannot be verified cannot be defaulted, recommended, or trusted with a sponsored moment.

Build one genuinely useful ambient behavior before buying any ambient media. A reorder skill, a fitment checker, a maintenance reminder. Something a user would summon on purpose. Utility is the visa that gets you into the environment legally; media alone gets you deported.

Assign an owner to the defaults battle now. Somebody in your organization should wake up thinking about how your brand becomes the unspoken fulfillment choice for the assistants your customers use. That person needs partnerships skills, technical fluency, and a budget line that does not compete with quarterly performance spend.

Ration your presence deliberately. Set an internal cap on unrequested ambient touches per user per month, and hold it even when the platform lets you exceed it. Treat the Presence Budget as real because the user's memory is real.

And retrain creative. Brief writers on moments, not messages: what is happening around the user, what would actually help, and what is the fewest words that help can take. The brands that learn to be quiet correctly will own the loudest decade in computing.

Cite this paper

Subeh, P. (2026). Ambient Advertising: Marketing Inside Always On Computing. The Future of Advertising Papers, No.26. https://www.pierresubeh.com/research/ambient-advertising-always-on-computing

No.25

The Attention Recession: Modeling the Shrinking Human Ad Supply

No.27

The Price of Interruption in a Post Feed World

More in Attention Economics