Amazon Just Quietly Built A $70 Billion Ad Business And Half Of You Are Still Pretending Meta Is The Only Game In Town
—two minutes into the post parade and people are still acting like the Kentucky Derby is just a horse race instead of one of the most lucrative single-day betting events in North America, but fine, let’s start there because today is a perfect example of how the same person who can read a 5-1 morning-line on Renegade and intuit what it means cannot read a $17.24 billion Q1 advertising number from Amazon and intuit what THAT means.
Amazon dropped earnings Wednesday. April 29. Read the release, the actual press release on aboutamazon.com, not the financial-media headline that just said “AWS beat.” Here’s what was buried in there. Amazon advertising services revenue grew 24% year over year to $17.24 billion in a single quarter. Trailing twelve-month advertising revenue is now over $70 billion. Andy Jassy said it on the call, almost as a throwaway: “Advertising grew to over $70 billion in TTM revenue.”
Seventy. Billion. Dollars. In a year. From advertising. From Amazon.
For comparison, that’s larger than the entire AWS business was in 2018. It’s larger than YouTube’s ad revenue. It’s roughly comparable to Meta’s quarterly ad revenue but it’s growing faster on a percentage basis in 2026 than Meta is. And almost nobody in the dropshipping ecosystem talks about it because Amazon advertising still has the reputation of being “that thing Amazon sellers do,” which it stopped being two years ago.
Meta’s Q1 came out Wednesday too. $55 billion in ad revenue, 33% growth, you’ve seen it. The headline writers all wrote the same article. What nobody connected was that Amazon and Meta both reporting blowout ad quarters in the same week, with Amazon advertising growing at 24% and Meta growing at 33%, on a base where Amazon is now a quarter the size of Meta in pure ad dollars but pulling ahead on certain ROAS metrics in certain categories. The shift is real. The duopoly is a triopoly now and the triopoly is reshaping in real time. If your entire customer acquisition strategy is “scale Meta, maybe try TikTok Shop,” you are operating with one and a half tools in a three-tool world.
Today happens to be Saturday May 2, 2026. The 152nd Kentucky Derby goes off at Churchill Downs at 6:57 PM Eastern, Renegade out of post position 1 as the morning-line favorite at around 5-1, NBC and Peacock carrying the broadcast. I bring this up because Kentucky Derby Day is one of the largest single-day spikes in sports betting volume in North America, and FanDuel and DraftKings have been blanketing every digital channel for two weeks running affiliate offers — $25 in bonuses with a $5 wager from FanDuel Racing, $400 in bonus bets from TwinSpires through CBS Sports. That ad spend is going somewhere. Most of it is going to Meta and Google. A growing slice of it, every quarter, is going to Amazon, because Amazon’s ad inventory is no longer just sponsored product placements — it’s Prime Video ads, Twitch ads, IMDb TV ads, Fire TV inventory, Alexa interactions. Amazon owns the largest TV-streaming OS in the country. Their advertising platform is buying inventory you don’t even associate with Amazon when you see it.
So why do dropshippers still ignore Amazon Ads as a customer acquisition channel for off-Amazon traffic? Because the YouTube content economy hasn’t gotten there yet. Because Sponsored Products got branded as “for Amazon sellers only” five years ago and the perception calcified. Because the actual Amazon DSP — the demand-side platform, the thing that buys those Prime Video and Fire TV impressions — has a steep learning curve, no self-serve tier in any meaningful sense for small advertisers, and historically required a managed service relationship.
That last one is changing. Quietly. Right now. Amazon DSP opened up self-serve access more broadly in late 2025. The minimum spend used to be $50,000 a month managed. Now you can get into self-serve DSP at meaningful but achievable thresholds — depending on your account rep and category, sometimes as low as $5,000 to $10,000 per month total committed budget — and run retargeting, lookalike, and contextual campaigns against Amazon’s first-party shopping data on platforms outside Amazon. That is the most underpriced traffic source in commerce right now and most operators in this space have never logged into it.
You with me? Because I want to be specific.
What’s actually working right this second on Amazon DSP for a Shopify-first dropship brand: a three-layer audience structure. Layer one is in-market audiences for your category — Amazon knows who has been browsing electric kettles or skincare or pet beds in the last 30 days, that’s data Meta literally cannot match because Meta doesn’t see purchase intent the way Amazon does. Layer two is competitor-conquesting audiences — Amazon will let you target users who have viewed specific competitor ASINs even if you don’t sell on Amazon. Layer three is your own pixel-based retargeting from Shopify, fed in via Amazon’s tag. You run video creative through Prime Video and Fire TV, drive to your Shopify storefront, attribute through Amazon Attribution which you can set up free even without selling on Amazon. CPMs are higher than Meta. CPAs in early tests across multiple verticals are lower because intent is so much higher. The math works.
This is the entire point everyone keeps missing. Meta is selling attention. Amazon is selling intent. Those are different products and they should command different prices and they should occupy different parts of your funnel. Treating them as competing top-of-funnel platforms is what gets you outbid by the brands who figured out the stack.
Now the thing everyone’s doing wrong this week, in the wake of the earnings.
Reading “Meta ad revenue up 33%” and concluding “I should pour more into Meta.” That is the reflex and it is wrong, or at least it’s incomplete. Meta’s ad revenue is up because (a) prices per impression went up 12% and (b) total impressions went up 19%. The price increase is a tax on every advertiser including you. The impression increase is mostly Reels inventory, which monetizes worse for direct response than Feed. So Meta’s revenue is up but the average advertiser’s CPA in direct response is rising in the same categories where Meta’s own ad load is the heaviest. Meta makes more money. You spend more for the same conversion. Those are not contradictory facts, they are the same fact described from two sides.
The right reaction to this earnings season is not “spend more on the platform that just got more expensive.” It’s “diversify into the channels that are growing without inflating CPMs at the same rate.” Amazon DSP. Reddit Ads, which has been quietly building out a real performance product. TikTok Shop affiliates, which I keep saying every week and people keep ignoring. Pinterest, which is having a quiet renaissance with Gen Alpha that no one is reporting on because Pinterest doesn’t issue earnings press releases that drive headlines.
This is also why things like https://whop.com/the-ecom-linked-advantage even exist — because the readers of Wall Street Journal financial coverage and the readers of dropshipping Twitter are reading entirely different versions of the same earnings report and most operators only see the version filtered through whatever bro on YouTube tells them to think.
Now the May Day rant, because we have to.
Yesterday — Friday May 1 — was May Day. Nearly 500 organizations coordinated more than 750 events nationwide under the “Workers Over Billionaires” banner, with the explicit message: “No School. No Work. No Shopping.” The National Education Association was a key organizer. Charlotte-Mecklenburg School District canceled school. Major rallies in New York, DC, Chicago, LA, Minneapolis, Seattle. Whether you agree with the politics or you don’t is genuinely irrelevant for our purposes. What matters is the commercial reality: a coordinated boycott-of-shopping campaign was happening on a Friday, the day Meta and Amazon ad systems both treat as a peak conversion day, and most direct-to-consumer brands ran their ad budgets exactly as they would on any other Friday because nobody updated their calendars.
I watched ad-spend data across a few categories yesterday. CPMs were normal. CTR was normal. Conversion rate dropped meaningfully in lifestyle, apparel, and home categories. Not catastrophically, but enough that anyone on tight margins had a worse-than-usual Friday and didn’t know why because the protest didn’t trend on the brand-side dashboards. The platforms don’t tell you “by the way, 750 events nationwide are explicitly asking your customers not to buy things today.” That intel comes from reading actual news, which most operators don’t do, because they’re watching dropshipping creators react to other dropshipping creators reacting to a leaked agency case study.
The lesson here is bigger than May Day specifically. It’s that the calendar matters more in 2026 than it ever did, because consumer attention has fragmented to the point where any of a dozen “events” — protests, awards shows, sports finals, viral moments, political news cycles — can suck conversion air out of a 24-hour window. Pulling spend on those days, or shifting it to evergreen retargeting, or just being aware enough to not blame your creative when the issue was the day, is the kind of operator-level discipline that separates a profitable account from one that’s hemorrhaging on autopilot.
So here’s the strategy, the actual implementable thing.
Build a “blackout calendar” for the next 90 days. Don’t outsource this. Sit down with a coffee, open a Google Doc, and write down every single date where you can identify a likely drop in commercial-conversion intent. Today: Kentucky Derby Day, sports betting and NBC eyeballs eating discretionary attention 4 PM through 8 PM Eastern. Cinco de Mayo on Tuesday (different conversion pattern depending on category — restaurants up, ecom often down). Mother’s Day on Sunday May 10, which is the opposite — peak conversion for gifts, peak competition. Memorial Day weekend May 23-25, traditional sale window, expect CPMs to spike 20-40% across paid social. NBA Finals starting in early June. Then the bigger global stuff — Father’s Day, Fourth of July, Prime Day in mid-July (which I’ll talk about in a second), back-to-school cycle.
For each date, decide in advance: are you spending more, the same, less, or shifting channels? Write the answer in the doc before the day arrives. The single biggest cause of bad ad performance I see in this industry is operators who let their automated bidding strategies run blindly through the calendar without making any active calendar-aware decisions, and then they tweak the campaigns reactively after the damage is done. Be proactive. The calendar is a known thing. Let the algorithm be smart about who to serve to. You be smart about WHEN.
About Prime Day. Amazon hasn’t officially announced 2026 dates as of today — historically it’s been mid-July, and given AWS just spent $44 billion in capex this quarter and the implied $200 billion full-year capex Andy Jassy is telegraphing, you can bet your kidneys that Prime Day 2026 will be the most aggressive Amazon Prime Day in history. They need to demonstrate to investors that the AI infrastructure spend is monetizing. They will pour creative resource into Prime Day. For ecom operators NOT on Amazon, Prime Day historically suppresses Shopify direct-response conversion rates by 8-15% across most non-grocery categories during the 48-hour window because attention concentrates. Plan for it now. If your Prime Day strategy is “we don’t sell on Amazon so we don’t care,” you are wrong. You’re losing customer attention whether or not you’re competing for the listing.
Now a thing most people don’t know about, since you read this far.
Amazon Sponsored Display, which is technically separate from DSP, lets you target shoppers off Amazon based on ASINs they’ve viewed on Amazon. So even if you’re not selling on Amazon, you can — through Sponsored Display in some account configurations — buy ad placements on Amazon’s network targeting people who looked at competitive products. The implementation is fiddly. The targeting is filtered through the lens of ASINs not keywords. But for a Shopify brand that competes with established Amazon products, this is one of the cleanest intent signals available anywhere in digital advertising. Most agencies won’t proactively suggest this because the management overhead doesn’t fit their fee structure and most in-house marketers don’t know it exists because their agency hasn’t told them. The information edge persists exactly because of these structural incentives, which is why things like https://whop.com/the-ecom-linked-advantage trade in private operator chatter — the consultants who could explain this stuff publicly are paid not to.
Apple’s earnings dropped Thursday after-hours, the day before yesterday. Tim Cook called the iPhone 17 lineup “the most popular in its history.” Mac mini and Mac Studio supply constraints will persist for several months according to the company. Why this matters for ecom operators: holiday season 2026 is going to be the first holiday where iPhone 17 is the dominant device in the wild, and Apple’s privacy framework continues to harden. App Tracking Transparency was 2021. The Private Relay rollout has been gradual. iOS 19 introduced more aggressive default opt-outs in 2025. The pixel-based attribution model is a fading thing. By Q4, your reliance on Meta’s reported numbers vs. actual incremental contribution is going to be the biggest gap most operators have ever seen, and the only fix is doing your own incrementality testing — geo holdouts, paid-vs-organic split tests, blackbox lifts — before the holiday season starts. Triple Whale, Northbeam, Rockerbox have post-purchase survey integrations that are imperfect but better than nothing. If you don’t have one configured by July you’re going to spend Q4 in a fog.
Prediction for the next 30 days, since I owe one.
Between now and the start of June you’re going to see at least one major Amazon advertising platform announcement positioning DSP and Sponsored Display more aggressively at SMB ecom advertisers. The earnings call essentially telegraphed it — when Jassy emphasizes “advertising grew to over $70 billion TTM” the way he did, that’s a flag to investors that this is the next AWS-level growth story, and Amazon’s pattern in those moments is to follow the financial signaling with a productization announcement designed to onboard the next tier of advertisers. They’re going to come for the Shopify-first ecom market. The early movers who are already running Amazon DSP through agency-of-record arrangements are going to be the ones who already have learning, already have audience data, already have creative assets warmed up. The folks who try to spin it up in October when Q4 panic hits are going to pay double.
Other 30-day calls, faster. TikTok Shop affiliate commission rates are going to drop in saturated categories (beauty, supplements) as enterprise brands flood in — I said this last week, I’m saying it again because the trend accelerated this week with Ralph Lauren and Olaplex onboarding. Reddit Ads will roll out at least one new commerce-specific feature; their sponsored product placement integration is in beta with select partners and full rollout is overdue. And at least one mid-sized Shopify app in the AI-content space will get acquired by either Klaviyo or Yotpo, because the consolidation cycle in martech is overdue and 2026 is when it happens.
Tools I see in healthy stacks right now: Triple Whale or Northbeam for attribution (pick one, the difference is marginal, the consistency matters), Motion or AdCreative.ai for static creative iteration but not video, Pencil or Arcads for early-stage video iteration, Stormy AI or similar for product-feed structuring, Gorgias or Re:amaze for support, Klaviyo for email/SMS still (no, ActiveCampaign is not better for ecom, fight me), Recharge or Skio for subscription if you have any subscribe-and-save SKUs. For Amazon advertising specifically: Pacvue and Skai are the heavyweight platforms, Perpetua is friendlier for smaller accounts, and the native Amazon advertising console has gotten dramatically better in the last six months and is genuinely usable for accounts under maybe $30k/mo in Amazon spend. I am not telling you which to use. I am telling you what’s in operator stacks. Do your own diligence.
What I will not endorse: any of the “AI agent that runs your store autonomously” pitches that have flooded the space since OpenAI’s last release cycle. They’re mostly demos. The technology is not there. It will be. It is not. Letting an LLM with browser access touch your ad accounts unsupervised in May 2026 is a fast way to lose a month’s profit in a weekend.
Last thing about today.
Renegade is the favorite at the Derby. Bob Baffert has two horses entered for the first time since his Medina Spirit suspension wrapped — Potente and Litmus Test. Whether either wins or not, the broader pattern of the day is what matters: a major sports event, a major sponsor scrum, several billion dollars in bets and ancillary commerce, and an entire shadow economy of affiliate marketing, prop bets, and second-screen content that most ecom operators consider entirely outside their world.
It isn’t outside your world. It’s the same world. Attention is attention. Inventory is inventory. The only difference between an Amazon DSP impression on Prime Video during a Premier League match and a Meta Reels impression at 3 AM is which platform is collecting the rent and what kind of audience signal is being applied. The operators who win 2026 are the ones who think about it as one big media market with different vendors selling different audiences at different prices, not as separate “channels” that each get their own siloed strategy meeting.
Quick recap. Amazon advertising hit $17.24B in Q1 with $70B TTM. Most dropshippers are ignoring it. They shouldn’t. DSP self-serve is the move. May Day yesterday meaningfully suppressed Friday conversion rates in some categories and most operators didn’t notice. Build a blackout calendar for the next 90 days. Apple’s iPhone 17 dominance plus continued tracking restrictions means your Q4 attribution is going to be a mess unless you start incrementality testing now. Prediction: Amazon will productize DSP for SMB ecom in the next 30 days. Tools that matter, tools that don’t, all listed above.
Renegade off the rail at 6:57 Eastern. Most of you reading this will watch the race and never connect that the same sponsorship dollars funding the broadcast are the same dollars competing for your customer’s attention tomorrow morning when she opens Instagram. They are. Act accordingly.
P.S. — https://whop.com/the-ecom-linked-advantage if anyone’s curious.