Amazon Cracks the Ad Duopoly: The Triopoly Era Begins

For twenty years, digital advertising has been run like a two-party political system: Google on one side of the aisle, Meta on the other, and every brand voter forced to pick a lane. Amazon just walked in as an independent with a war chest, and the polls have moved. Q2 ad revenue of $19.8 billion, up 26% year over year, is not a protest candidate's numbers. That's a governing majority in waiting.

The interesting bit isn't that Amazon is big. It's that the shape of the race has changed, and most performance marketing stacks were built for the old two-party map.

The Numbers

Start with the raw scoreboard. Alphabet booked $81.6 billion in advertising revenue in the quarter. Meta pulled $60.8 billion. Amazon came in at $19.8 billion, as Storyboard18 reported, growing at 26% year over year while the two incumbents grow off a much larger base.

The old shorthand was the $160-billion duopoly. That number now needs an asterisk the size of a Prime Day banner. Amazon isn't matching Google or Meta in absolute dollars, and won't this year or next. But growth rate is the tell. When the smallest of three players is expanding fastest, the mix shifts every quarter without anyone voting for it.

Look at what Meta squeezed out in the same period: ad impressions up 14%, average prices up 12%. Those are healthy numbers for a company reaching 3.6 billion people a day across Facebook, Instagram, WhatsApp and Threads. But they're the numbers of a mature ad system running the dials, not a business discovering new inventory. Meta is optimising a machine. Amazon is still building one.

The other piece of the picture is what Amazon is actually selling. Sponsored product listings. Search placements on the retail surface. Prime Video ads. Live sports sponsorships. That's four distinct inventory types, each with its own auction mechanics, each capable of scaling independently. Anyone who has run a media mix model knows what happens when a new channel with its own attribution signal shows up: budget rebalances toward the channel where the conversion event is closest to the exposure event. Amazon owns the checkout. That's not a small structural advantage.

Zoom out and the retail media wave is bigger than one company. Walmart Connect in the US. Flipkart Ads, Blinkit Ads and Zepto in India. Every serious commerce platform is now a media platform, or building the team to become one. The consumer-goods brands, beauty houses, electronics OEMs and multi-category retailers named in the reporting are the ones writing the cheques. When the CPG buyer moves budget, the whole ad ecosystem feels it within two quarters.

What's Actually New

The lazy read on this story is "Amazon big, Google and Meta scared". The lazy read misses the interesting bit, which is what Amazon expanded during the quarter: Ads Agent, an AI-powered platform that reportedly gets brands from brief to live campaign in minutes instead of hours. Advertisers using it are seeing lower customer-acquisition costs.

Read that sentence twice if you run performance marketing. The pitch isn't better targeting or more inventory. The pitch is time-to-live and CAC. Those are the two metrics that actually govern where a mid-market brand's next dollar goes. Anyone who has watched a media planner spend three days building a campaign structure that could have been generated from a product feed knows the pain being solved here.

Google is embedding AI into search. Meta is using AI to sharpen targeting and recommendations. Amazon is automating the campaign planning, audience targeting and media buying itself. Those are three different theories of where AI creates the biggest lift in the ad stack. Google's bet is on the query side. Meta's bet is on the signal side. Amazon's bet is on the operator side, replacing the human who used to sit at the console.

The operator-side bet is the one I'd watch closely. Query-side AI improves an experience you already have. Signal-side AI improves outcomes on inventory you already buy. Operator-side AI removes labour cost from running campaigns, which changes who can afford to be a serious advertiser. A DTC brand that couldn't justify a full-time paid media hire can now run structured campaigns from a product catalogue. That's a new tier of demand entering the auction, and it competes with the same eCPMs the incumbents rely on.

The other genuinely new thing is closed-loop attribution as table stakes. For fifteen years, marketers have been buying Google and Meta impressions and stitching conversions together with pixels, server-side events, and, more recently, the Conversions API. Retail media collapses that stack. The exposure and the purchase happen inside the same walled garden. There's no cross-domain identity problem to solve. In a world where the Privacy Sandbox is slowly reshaping what third-party attribution even looks like, that architectural advantage compounds.

What's Priced In for Performance Marketing

Most senior performance teams already assume Amazon is a top-three channel. That part is priced in. What's not priced in, and this is where I see teams get caught out, is the second-order effect on media mix modelling and creative production.

Media mix models built between 2019 and 2024 treat Google and Meta as the two dominant paid channels and everything else as a rounding error. Rebuilding those models with retail media as a first-class citizen isn't a config change. It's a re-instrumentation project. You need clean feeds from Amazon, Flipkart, Walmart Connect and every retail network your category cares about, normalised into a schema that lets you compare incremental ROAS across surfaces. Anyone who has tried to reconcile Amazon's reporting with a third-party MMP knows this is the boring bit that eats quarters.

The creative production side is where I'd argue the shift is genuinely underestimated. Google and Meta ads are built for context. Retail media ads are built for a shopper who is already in cart-mode. That's a different creative brief, a different asset library, and different testing cadence. The performance teams treating Amazon like "just another paid channel" and shipping the same hero video are the ones who'll see flat ROAS while their more disciplined competitors compound.

Also priced in: AI-driven campaign automation from all three players. Everyone knows Ads Agent, Performance Max and Advantage+ exist. What's not priced in is the operational reality that once campaign structure is automated by the platform, the human lever moves to feed quality, product taxonomy, and creative variance. The winners will be the teams that treat their product catalogue like production infrastructure.

Contrarian View

Here's where I'd push back on the triopoly narrative. Calling Amazon the third pillar of digital advertising treats "advertising" as one market. It isn't. Amazon owns the bottom of the funnel. Meta owns the middle. Google owns both extremes, the top through YouTube and the bottom through search intent. These aren't three companies competing for the same dollar. They're three companies competing for three different jobs on the same media plan.

The triopoly framing also flatters Amazon. $19.8 billion is roughly a quarter of Google's number. Growth rates converge as the base grows. If Amazon's ad business slows to Meta-like growth in 2027 or 2028, the gap to the top two stays enormous for a long time. The story of a duopoly ending is a great headline. The reality might be a duopoly with an important supplier at the bottom of the funnel that everyone has to buy from but nobody has to shift their strategy around.

And a quiet worry for Amazon's ad business specifically: it's tied to the health of Amazon's retail business. If commerce growth softens, the ad surface softens with it. Google's and Meta's ad businesses aren't hostage to the performance of a single retail flywheel. That's a structural fragility worth remembering before writing anyone's obituary.

Key Takeaways

• The numbers matter less than the growth rate. Amazon's $19.8 billion trails Alphabet's $81.6 billion and Meta's $60.8 billion, but 26% year-over-year growth on a base that size reshapes the mix every quarter.
• Ads Agent is the story inside the story. Automating campaign planning and media buying, not just targeting, is a different theory of AI in advertising, and it opens the auction to a new tier of smaller advertisers.
• Closed-loop attribution is a structural moat. Retail media platforms bypass the cross-domain identity problem that Google and Meta increasingly have to engineer around.
• Rebuild the media mix model. Treating Amazon, Walmart Connect, Flipkart Ads, Blinkit Ads and Zepto as one-off channels rather than first-class inputs is where teams will lose ground over the next four quarters.
• Don't overread the triopoly headline. These three companies increasingly own different jobs on the funnel. The interesting competition is between Amazon and every other retail media network, not between Amazon and Google.

To go back to the two-party map: the independent candidate didn't win the election. But the coalition has shifted, the district lines have moved, and anyone still running a campaign strategy from the 2019 playbook is going to lose seats they thought were safe.