AI Vidia builds the ai creative Amazon DSP benchmarks in this report from Amazon's own creative specifications and Tinuiti's Q2 2026 ad spend data, and the short answer is that Amazon DSP is now the fastest growing ad line in Amazon's stack while most brands still feed it a single static product photo. Amazon DSP spend grew 67 percent year over year in the second quarter of 2026, up from 41 percent in the first quarter, according to Tinuiti's benchmark panel of more than USD 4 billion in managed ad spend. Amazon's own data shows AI enhanced creative lifts click through rate by 14 percent on average. The gap between what Amazon's Responsive eCommerce Creative engine can render and what most advertisers actually submit to it is where this benchmark starts.
Why Amazon DSP creative is the bottleneck nobody is watching
67%DSP SPEND GROWTH, YOY Q2 2026
+14%CTR LIFT, AI ENHANCED CREATIVE
20ASINS PER REC CREATIVE
10SUPPORTED CREATIVE SIZES
Amazon DSP spend does not just outpace the rest of Amazon's ad stack, it is separating from it. Sponsored Products grew 38 percent year over year in Q2 2026 and Sponsored Brands grew 20 percent, both real acceleration, but Amazon DSP's 67 percent growth is the fastest line in the entire report and the fastest Amazon DSP has grown in nearly five years. Every one of those dollars still routes through the same creative engine, Responsive eCommerce Creative, known as REC, and REC only reaches its full 10 supported placement sizes and up to 20 ASINs per creative when it receives a complete input: a square image, a tall image, and a wide image at three different aspect ratios, not the single default product photo most sellers upload once and never touch again. Call the gap between REC's actual capacity and what most brands feed it the REC ceiling, the point where a brand's Amazon DSP performance stops being a bidding problem and starts being a creative supply problem.
The cost of hitting the REC ceiling is not hypothetical. Amazon's own platform data shows that switching on AI enhanced creative, letting Amazon's models select and optimize images, headlines, and layouts, lifts click through rate by 14 percent on average. That lever sits idle on any creative that never got built with a real image set to optimize across. Amazon's creative rules add a second, sharper cost: no calls to action inside the image, no pricing or savings claims in the creative, no white or transparent backgrounds, a 10 word cap on in image copy. Creative built for a Meta feed or a TikTok For You page routinely breaks two or three of those rules on its own, so teams either burn review cycles fixing rejections or quietly ship the one image variant that already happens to clear the policy bar. Neither path uses the size range, the ASIN rotation, or the AI optimization Amazon actually built.
Amazon DSP is the fastest accelerating line in the Amazon ad stackQ2 2026 YoY ad spend growth by Amazon ad product
The pattern in the chart above is the reason creative supply now matters more on Amazon than it did two years ago. A channel growing 20 to 38 percent a year can absorb a thin creative set without the gap showing up in the numbers. A channel growing 67 percent a year, on a base that was already accelerating, turns the same thin creative set into a widening tax on every incremental dollar, because more spend is chasing placements the creative was never built to fill.
The Amazon DSP creative format matrix
Amazon DSP is not one creative format, it is four, and each one carries different specs, different placements, and a different bar for what counts as ready to ship. The table below is the matrix AI Vidia uses to brief a DSP creative set before a single image gets generated.
Format
Where it runs
Dimensions
File type and limit
Creative input required
Responsive eCommerce Creative, static
On Amazon, third party sites, Echo Show, Fire TV
10 supported sizes, 300x50 up to 1920x1200
JPEG or PNG, up to 5MB per custom image
Default product image or three custom images at 1:1, 9:16, and 1.91:1
REC with video
Third party publisher sites, US only
16:9
MP4, up to 500MB
Brand video, sourced footage, or AI generated video
Online video, OLV
Web and app, click through enabled
1920x1080 horizontal or 1080x1920 vertical
MP4, 16:9 minimum frame
Full video cut, no REC fallback if missing
Streaming TV, OTT and CTV
Fire TV, connected TV, Prime Video, Freevee
1920x1080, 16:9
MP4, broadcast quality
Full video cut, no click through, no text heavy end cards
Three rows in that matrix decide most DSP media plans. Responsive eCommerce Creative is the format nearly every advertiser touches first, and it is also the format most likely to ship with only a default image, since Amazon will technically accept that and still serve an ad. REC with video is the newer, higher ceiling option: it now runs directly inside the banner format on third party sites, in the US, and it accepts AI generated video as a valid input, which puts it well within reach of a studio production line already shipping AI video for other channels. Streaming TV carries the strictest bar of the four, since Fire TV and Prime Video inventory expects a broadcast quality cut with no click through and no text heavy end card, closer to a 15 second TV spot than a performance banner.
Amazon DSP is pulling away from Sponsored Products, not just outgrowing itYoY ad spend growth, Q1 2026 vs Q2 2026
Tinuiti Q1 2026 and Q2 2026 Digital Ads Benchmark Reports
The acceleration itself is the planning signal. Amazon DSP grew 41 percent year over year in the first quarter of 2026, already its fastest pace in nearly five years, then accelerated another 26 points to 67 percent in the second quarter, roughly double the acceleration Sponsored Products posted over the same two quarters. A creative set that was adequate against a 41 percent growth line is running short six months later, not because the creative got worse, but because the budget behind it grew two thirds larger while the format matrix stayed the same four rows.
Want a structured plan for your AI creative pipeline? 20-minute call, no pitch deck.
The Completeness Audit is the strategic model AI Vidia runs before recommending a media plan on any account moving budget into Amazon DSP. Five checks, run against the format matrix above, and the output is a creative production brief instead of a guess.
Step 1. Inventory what is actually live in REC today. Pull every active REC creative and check whether it carries a default product image only or a full three image custom set. Most accounts that have never audited this find the majority of live creative is default image only, which caps eligible sizing and skips the AI optimization lever entirely.
Step 2. Confirm AI enhancement is switched on. The setting that switches on Amazon's 14 percent average CTR lift is opt in, not default. An account can have a complete image set and still leave that lift on the table if the AI enhancement toggle was never checked during setup.
Step 3. Audit ASIN coverage per creative. REC supports up to 20 ASINs per creative with automatic rotation and out of stock protection. Most brands run one to three ASINs per creative, which forfeits both the rotation benefit and the stock out safeguard that stops budget from serving a sold out product.
Step 4. Map creative completeness to placement eligibility. Cross reference the format matrix against what is actually submitted. A default image only creative is not eligible for the same placement and size breadth as a full custom set, so this step turns a vague reach problem into a specific list of missing assets.
Step 5. Set a refresh cadence tied to the spend curve, not the calendar. A creative set sized for last quarter's budget is undersized for a channel accelerating 26 points in a single quarter. Tie the refresh trigger to spend growth thresholds, not a fixed monthly or quarterly date.
Kevin's take
That is why the Completeness Audit above starts with an inventory, not a bid recommendation. A media buyer can tighten targeting and adjust bids all quarter and still be capped by a creative set that was never built to use the size range, the ASIN rotation, or the AI optimization Amazon already ships for free. Fix the input first and the bidding conversation gets a lot shorter.
Framework 2: The DSP Creative Production Loop
The Production Loop is the tactical build plan AI Vidia runs to close the gaps the Completeness Audit finds, sized to Amazon's own format matrix rather than a generic content calendar.
Step 1. Brief against three ratios and one video cut, not one hero shot. Every REC brief specifies a 1:1 square, a 9:16 tall, and a 1.91:1 wide image, plus a 16:9 video cut for REC video or OLV. A single hero photograph cannot be cropped into all three without breaking Amazon's composition rules, so the brief has to plan for three compositions from the start.
Step 2. Generate against Amazon's creative rules, not general ad creative rules. No call to action text or button in the image, no pricing or savings claims, no white or transparent background, no more than 10 words of in image copy. Building to these constraints during generation avoids the rejection and resubmit cycle that eats a full review turnaround per fix.
Step 3. Run a brand lock and policy pass before submission. A creative director checks palette, product framing, and logo usage against the brand lock, then checks the same asset against Amazon's acceptance policy before it goes anywhere near the DSP console. Catching a policy break here costs minutes, catching it after Amazon's review costs a full cycle.
Step 4. Submit with AI enhancement opted in and monitor the size report. Amazon's DSP report center shows performance by placement size once a creative goes live. That report is the fastest way to confirm the full image set is actually earning eligibility across all 10 supported sizes, not just the two or three it shipped with previously.
Step 5. Rotate the set on a spend triggered cadence. Instead of a fixed monthly refresh, tie the next production cycle to a spend growth threshold, for example a 20 percent quarter over quarter increase in DSP budget. That keeps the creative set sized to the channel it is actually running in, not the channel it was running in last quarter.
Proof from the production line behind this benchmark
AI Vidia has not published an Amazon DSP case study yet, the format matrix and the two frameworks above are new to this report. What is proven, and what the DSP Production Loop is built on, is the same weekly production discipline AI Vidia runs for paid social clients today. For Andy Okay, the AI Vidia team ships 50 new AI ad creatives a week, with 1,000 or more AI ads produced to date across statics, UGC style video, and story ads. For IndianBites, documented in the IndianBites case study, the AI Vidia team shipped 142 AI ads in 11 weeks on a weekly 12 variant batch cadence, and the account held a 2.4x ROAS on its winning cohorts. A creative director reviews every asset before delivery on both accounts, the same review gate the DSP Production Loop above puts in front of Amazon's policy check.
Amazon did not build a 20 ASIN, three ratio, AI enhanced creative engine so brands could keep feeding it one product photo.
The mechanism that makes 50 variants a week possible for Andy Okay on Meta is the same mechanism a DSP creative set needs: a brief that specifies every required composition up front, generation against a fixed rule set, and a creative director gate before anything ships. Swap Meta's placement rules for Amazon's REC composition and policy rules and the production loop above is what that mechanism looks like pointed at Amazon DSP.
When each option wins
Stay with a default image only REC creative when Amazon DSP is still a test budget, under roughly USD 2,000 a month, and speed to launch matters more than size coverage. Below that spend level the REC ceiling costs less in absolute terms than a full production pass, and the tradeoff is rational.
Build the full three ratio custom image set once DSP becomes a standing budget line and the team wants eligibility across all 10 REC placement sizes. This is the point where the Completeness Audit typically finds the widest gap between what is live and what Amazon supports, and closing it is a design task, not a media buying task.
Add REC video and OLV once DSP spend is riding the kind of growth curve the chart above shows, and the account wants eligibility for Prime Video and third party publisher video inventory alongside static placements. This is also the point where AI generated video, already a valid REC input per Amazon's own specs, becomes the fastest way to keep a video cut current without a new shoot every quarter.
Bring in a dedicated creative production line, the model AI Vidia runs through AI Vidia's AI image ad service and AI Vidia's AI video ad service, once the team cannot keep the REC image set, REC video, OLV, and Sponsored Brands video all current on a spend triggered cadence at the same time. That is a multi format supply problem, and it is the same problem the Production Loop above already solves weekly for Meta and TikTok accounts.
The next step
The fastest way to find out where your account sits against the REC ceiling is a 30 minute scoping call. The AI Vidia team will run the Completeness Audit against your live Amazon DSP creative, map it to the format matrix above, and return a production brief sized to your current spend curve. This works well for ecommerce brands whose Amazon DSP budget is scaling faster than their creative line can keep up with, whether or not paid social is already in the mix. Book the call at AI Vidia's booking page, or read the companion benchmark on AI Vidia's CPM benchmark report for Meta and TikTok for the paid social side of the same production line.
Frequently asked questions
01What is the AI creative Amazon DSP benchmark for ad spend growth in 2026?
Amazon DSP spend grew 67 percent year over year in the second quarter of 2026, up from 41 percent in the first quarter, according to Tinuiti's Q2 2026 Digital Ads Benchmark Report. That makes DSP the fastest growing line in Amazon's entire ad stack, ahead of Sponsored Products at 38 percent and Sponsored Brands at 20 percent. The acceleration means a creative set sized for the first quarter of 2026 is already undersized six months later. Treat the benchmark as a planning signal for creative capacity, not just a media spend statistic.
02What creative does Amazon Responsive eCommerce Creative actually require?
Responsive eCommerce Creative, REC, accepts either a single default product image or a full custom set of three images at three aspect ratios: a 1:1 square, a 9:16 tall image, and a 1.91:1 wide image. REC also supports an optional video cut at 16:9, currently available on third party publisher sites in the US. Amazon renders whichever inputs it receives into up to 10 supported placement sizes automatically, but a default image only submission does not reach the same size and placement breadth as a full custom set. Up to 20 ASINs can be attached to one REC creative, with automatic rotation and out of stock protection built in.
03Does AI generated creative actually improve Amazon DSP click through rate?
Amazon's own platform data shows that switching on AI enhanced creative, which lets Amazon's models optimize images, headlines, and layouts automatically, produces a 14 percent average click through rate lift. That figure comes from Amazon Ads internal data published on its own creative specification page, not a third party estimate. The lift only applies to creative that has enough input to optimize across, so a single static image with AI enhancement switched on will not see the same gain as a full three image custom set. AI generated video and imagery are both accepted as valid REC inputs, which is what makes a studio production line a practical way to keep that input complete.
04How is Amazon DSP creative different from Meta or TikTok ad creative?
Amazon DSP creative has to satisfy Amazon's own composition and policy rules, which block calls to action inside the image, pricing or savings claims, white or transparent backgrounds, and more than 10 words of in image copy. Creative built first for a Meta feed or a TikTok For You page routinely breaks two or three of those rules, since those platforms allow or expect exactly the elements Amazon blocks. Amazon DSP also spans four distinct formats, REC static, REC video, online video, and streaming TV, each with its own dimensions and file specs, where a single Meta placement usually needs one or two creative sizes. Treat an Amazon DSP brief as its own production line rather than a resize pass on existing social creative.
05What does AI Vidia prove about creative production speed, and does it apply to Amazon DSP?
AI Vidia has not published an Amazon DSP case study, so this benchmark does not claim a DSP specific result. What is documented is the production model behind the frameworks in this report: 50 new AI ad creatives a week for Andy Okay, with 1,000 or more AI ads produced to date, and 142 AI ads in 11 weeks for IndianBites at a 2.4x ROAS on winning cohorts, both reviewed by a creative director before delivery. The same weekly brief, generate, review discipline that produces those numbers on Meta and TikTok is what the DSP Creative Production Loop in this report applies to Amazon's four format matrix instead. The production model transfers; the specific Amazon DSP performance numbers, once AI Vidia has a named case to report, do not yet exist.
06When should a brand upgrade from a default REC image to a full Amazon DSP creative set?
A default image only REC creative is a reasonable choice only while Amazon DSP is a test budget, roughly under USD 2,000 a month, where speed to launch matters more than size and placement coverage. Once DSP becomes a standing budget line, building the full three ratio custom image set closes most of the eligibility gap the Completeness Audit in this report is built to find. Adding REC video, online video, and streaming TV creative makes sense once DSP spend is on a sustained growth curve like the 67 percent year over year rate this report benchmarks. Below the test budget threshold, the reach a full creative set buys usually costs more than the REC ceiling itself.
