The week ending June 6 compressed several years of directional change into a handful of announcements. OpenAI activated its advertising pilot in the United Kingdom this morning, making it the first market outside North America, Australia, and New Zealand to receive ChatGPT ads. On the same two-day stretch, DoorDash repositioned itself as a full commerce media platform, Pinterest committed $4 billion to Amazon Web Services through 2031, Roblox opened advertising to children under 13 for the first time, and a Blackstone-backed mobile ad company called Liftoff raised $437 million on the Nasdaq. None of these stories exists in isolation. They are, taken together, a snapshot of where ad spending is flowing, and what infrastructure is being built to handle it.

OpenAI takes ChatGPT ads to the UK - and further

The most consequential move of the day belongs to OpenAI. According to PPC Land, OpenAI's VP of Monetization, Benji Shomair, confirmed on June 6, 2026 via LinkedIn that the ChatGPT advertising pilot is now active in the United Kingdom. The announcement is significant for several reasons. The UK is the first European market to go live, and it arrives alongside confirmation that Japan, South Korea, Brazil, and Mexico will follow in the coming weeks.

The pilot began in the United States on February 9, 2026, with a small roster of premium advertisers and a minimum spend requirement of $200,000 to $250,000. That threshold has since been eliminated entirely. On May 5, 2026, OpenAI opened a self-serve Ads Manager beta to all US businesses, adding cost-per-click bidding at a recommended starting rate of $3 to $5 per click alongside the existing CPM model. Daily budget controls, geographic targeting at DMA and ZIP code level, and dynamic call-to-action options followed on May 22. The UK launch inherits all of those features.

The subscription boundaries OpenAI established at launch remain intact. Ads reach users on free and Go tiers only. Plus, Pro, and Enterprise subscribers will not see any advertising. The Go tier costs $8 per month globally; according to PPC Land, roughly 85 percent of ChatGPT users were eligible to see ads as of March 26, yet fewer than 20 percent encountered one on any given day during the pilot phase. That deliberate restraint on ad load has been one of the levers OpenAI is monitoring alongside three formal metrics: trust, dismissal rates, and relevance. When OpenAI published a progress update on March 26, it reported no negative impact on consumer trust, low dismissal rates, and ongoing relevance improvements.

The financial trajectory has been steep. According to PPC Land, the US pilot crossed $100 million in annualised revenue within six weeks of its February launch. That figure arrived when the minimum spend was still $50,000; since the barrier was removed and CPC bidding introduced, the accessible advertiser base has grown considerably. Criteo, the first formal ad tech partner confirmed in March, has published data from a sample of 500 US retailers showing that users referred from ChatGPT convert at approximately 1.5 times the rate of other referral channels. Technology partners now include Adobe, Kargo, Pacvue, and StackAdapt. Holding companies Dentsu, Omnicom, Publicis, and WPP are also supporting campaigns through the platform.

The UK launch introduces a different regulatory context. The Advertising Standards Authority is an active regulator. British consumers have demonstrated sustained attentiveness to data practices since GDPR came into force in 2018. OpenAI's stated advertising principles - answer independence, conversation privacy, long-term value, and user choice - will be tested against that backdrop. The answer independence principle, which commits the company to ensuring ads do not influence ChatGPT's responses, is particularly load-bearing in a market accustomed to scrutinising the line between editorial and commercial content.

For the broader advertising industry, the UK activation closes an important question. Whether OpenAI's ad-load restraint and consent mechanics travel from the US to privacy-sensitive markets is now a live experiment rather than a hypothetical. MediaPost reported on June 2 that conversion-optimised campaigns, where advertisers pay only when a user completes a purchase after clicking an ad, were expected to roll out in early June. If that timeline holds, the UK will be among the first markets to receive a fully performance-measurable ChatGPT ad product.

DoorDash makes the case for intent-based commerce media

One day before OpenAI's UK announcement, DoorDash spent June 4 recasting its advertising unit as something larger. According to PPC Land, the company now describes itself as a full-stack commerce media platform serving more than 400,000 advertisers across DoorDash, Wolt, and Deliveroo. Six product changes were announced simultaneously: a new homepage format called Spotlight, expanded global offsite reach through Symbiosys, a LiveRamp clean room for privacy-safe measurement, enhanced Smart Campaigns supporting buy-one-get-one promotions, and auto-bidding with a minimum return-on-ad-spend floor.

The 400,000 advertiser figure warrants context. It counts the full long tail of restaurants and grocers that use the platform, not only the brand partners that competing networks typically report. That methodology makes direct comparison with Amazon, Criteo, or Instacart awkward. What the number does communicate is the scale of the inventory base and the breadth of the addressable market for DoorDash's ad tools.

Spotlight, the headline format, occupies the homepage rather than search results. It targets discovery, the moment before a user has expressed explicit purchase intent, and DoorDash says early testing shows click-through rates twice those of standard banner formats. More pointedly, the company reports that first-time customers account for over 20 percent of restaurant sales and over 36 percent of CPG brand sales through the format. Toby Espinosa, VP of Ads at DoorDash, framed the proposition plainly: "Consumers come to DoorDash ready to buy, and that's a fundamentally different opportunity for advertisers than most platforms can offer."

The offsite component runs through Symbiosys, which DoorDash acquired in 2025. Symbiosys extends DoorDash's first-party audience data into search, social, and display campaigns across the Americas, Europe, and the Asia-Pacific region. Magnum Ice Cream Company was the first brand to activate Symbiosys' full social channel suite with DoorDash data; the campaign produced an 85 percent rise in new consumers against the prior period, though that is a single-advertiser result rather than a platform average.

Measurement runs through a LiveRamp clean room, where advertiser and DoorDash first-party data are matched without either side accessing the other's raw records. The aggregate outcome reported across campaigns analysed this way: over 80 percent of consumers reached were new to the advertiser's existing customer base. A national restaurant chain found 81 percent of its DoorDash engagers were absent from its own customer data entirely. Uber reached a comparable arrangement through its own LiveRamp deployment in December 2025; the structural parallel between two delivery platforms building privacy-centric measurement infrastructure is not coincidental. Both hold first-party transaction data that advertisers cannot access directly, and clean room technology is how that data becomes attributable without exposing individual records.

The automation additions are technically modest but operationally relevant. The new auto-bidding option, which more than 95 percent of test campaigns met their minimum ROAS floor in February, reduces the manual work required from advertisers managing large restaurant or CPG portfolios. Smart Campaigns' support for buy-one-get-one promotions extends a format that convenience and grocery advertisers use frequently.

Pinterest bets $4 billion on the AI infrastructure beneath its ads

The computing costs behind visual discovery came into focus on June 4, when Pinterest announced a $4 billion commitment to Amazon Web Services through 2031, the largest infrastructure deal in the company's history. MediaPost covered the deal on June 5, describing it as intended to accelerate Pinterest's AI roadmap and modernise the infrastructure powering its global visual search platform. The agreement formalises a relationship dating to 2010, but it adds specificity about hardware that the prior relationship lacked.

Pinterest will train its large language and vision-language models on AWS Trainium, Amazon's custom machine-learning silicon, and handle inference on AWS Graviton, the Arm-based processor that already runs roughly a third of Pinterest's compute. Training and inference are distinct engineering tasks at different cost structures; routing both through Amazon's proprietary chips reduces Pinterest's dependence on third-party graphics processors and, both companies argue, lowers the per-query cost at a scale where small efficiencies multiply.

The numbers behind the deal are worth examining. Pinterest reported 631 million monthly active users and $1.008 billion in revenue for Q1 2026, an 18 percent year-on-year increase. According to PPC Land, its Performance+ automated campaign suite - the ad product most directly dependent on the recommendation infrastructure the AWS deal is meant to improve - drove 30 percent of lower-funnel revenue by Q1 2026. Pinterest CTO Matt Madrigal described the arrangement as providing "the compute flexibility, hardware optionality, and infrastructure efficiency to accelerate our AI vision." The underlying technical structure is the Taste Graph, a multimodal AI system trained on billions of images, videos, and text that maps connections between user interests and content to generate personalised discovery feeds. The $4 billion is, in concrete terms, the bill for keeping that system current at the scale it now operates.

The timing also reflects a January 2026 restructuring that cut roughly 780 positions while redirecting resources toward AI. That combination of headcount reduction and infrastructure investment is visible across several large platforms simultaneously: the overall pattern is that the human labour of content curation and campaign management is being replaced by machine systems that are themselves expensive at scale.

Roblox opens to under-13s - but closes the programmatic door

On the same day DoorDash was broadening automation, Roblox was deliberately limiting it. According to AdExchanger, Roblox named SuperAwesome, a youth-focused ad marketplace, as its exclusive advertising partner for users under the age of 13 on June 4, 2026. This is the first time Roblox has made the under-13 audience available to brands in any form.

The constraints are substantial. There will be no programmatic buying for this cohort at all. Ads run as direct deals only, confined to home-screen banners and certain in-game billboards featuring static or video creative. Rewarded video – the format that incentivises continued platform use - is excluded. Roblox tied the decision to recent safety work, including new age verification and the introduction of two new user tiers: Roblox Kids for ages 5 to 8, and Roblox Select for ages 9 to 15. Those tiers give advertisers the structural separation to target campaigns by age group without relying on behavioral data or one-to-one personalised targeting, which are prohibited for the under-13 segment. Contextual targeting - based on the type of game in which an ad appears - and generalised demographic signals such as country, gender, and age range are the available levers.

SuperAwesome CEO Kate O'Loughlin described the classification work required to make contextual targeting functional at Roblox's scale. The platform hosts tens of millions of user-generated experiences, with roughly 15,000 new games added daily. SuperAwesome is responsible for categorising that inventory not only by genre but by gameplay style and game mechanics - a more granular taxonomy than most contextual systems apply.

The commercial logic is apparent even at the surface. Roblox had previously flagged by the Children's Advertising Review Unit in 2023 for failing to clearly label all branded experiences as advertisements. It introduced standardised labelling in May 2026. The combination of corrected labelling and a new, tightly controlled direct-deal channel represents a compliance-first entry into under-13 monetisation rather than an audience-reach play. SuperAwesome will collect a cut of revenue from these campaigns alongside Roblox's existing revenue share with game publishers.

Roblox's existing programmatic partnerships with PubMatic and Google Ad Manager are unaffected; the exclusivity agreement with SuperAwesome covers only the under-13 segment, which those partnerships do not touch.

Liftoff's IPO tests how investors now price AI in ad tech

The week's financial marker came from Liftoff Mobile, the Blackstone-backed mobile ad platform that raised $437 million on the Nasdaq at $23 per share. According to Digiday on June 5, the listing is the first notable ad tech IPO since MNTN's debut in May 2025 and therefore the most direct recent test of whether public investors are willing to back independent ad tech companies again.

The backdrop is mixed. Liftoff originally targeted a much larger raise - roughly $711 million at a $5.2 billion market cap - before withdrawing the filing in February when broader software stocks sold off sharply. The company returned with a more conservative pitch. The Q1 2026 numbers it brought to market are credible by most standards: $205.6 million in revenue, up from $149 million in the prior year period, driven primarily by Cortex, its neural network-powered prediction model for in-app bidding. Net income reached $49.3 million. Adjusted EBITDA hit $120.1 million, producing a 58 percent margin for the quarter - the figure most likely to attract sustained investor attention.

Liftoff sits in a sector where AppLovin has established an outlier benchmark: $1.84 billion in Q1 revenue with an 85 percent adjusted EBITDA margin. Liftoff's economics do not match that level, but they place it toward the higher-quality end of the public ad tech cohort. Both companies operate in the mobile app economy, and both position neural network-powered bidding models as their primary competitive asset. Liftoff's S-1 describes Cortex as capable of processing far more data and running more tests than conventional bidding systems, a claim that investors in the AppLovin era have learned to take seriously.

The company is the product of Blackstone's 2019 acquisition of Vungle, followed by a merger with the Liftoff platform the firm also held. That private equity assembly suggests a relatively compressed timeline for the public market debut, which is common in ad tech but raises questions about whether the combined entity has had sufficient time to build the kind of durable proprietary data advantages that AppLovin spent years constructing.

Search Engine Roundtable noted on June 5 that Google Search Console's AI performance reports were rolling out to a limited set of users at the same time the Liftoff listing landed. The parallel is coincidental but illustrative: measurement infrastructure is being built simultaneously at the platform level, the ad tech level, and the search level, each driven by the same pressure to demonstrate that AI-powered systems produce outcomes that justify their cost.

Google Search profiles and the publisher distribution question

The week's quieter but structurally significant development arrived on June 4, when Google formally launched Search profiles in the United States. According to PPC Land, the feature gives publishers and creators a claimable page that aggregates articles, videos, and social posts, and ties directly to the follow functionality that determines Discover feed distribution.

The follow mechanism is the operative part. When a user follows a publisher from their Search profile, Google increases the likelihood that publisher's content will appear in their Discover feed. The practical consequence is significant because, as PPC Land documented in December 2025, web search traffic to news publishers fell from 51 percent to 27 percent between 2023 and 2025, while Discover's share of Google referrals to news sites climbed to 67.51 percent over the same period. Discover access has effectively become a critical variable for publishing businesses, and Search profiles create a direct pathway into it.

Eligibility at launch is restricted to publishers and creators with a minimum following on at least one major platform: 300,000 on TikTok; 100,000 on YouTube, Instagram, or X. The feature is US-only at launch, with international expansion described as forthcoming. Search Engine Roundtable confirmed on June 5 that the profiles are officially live, describing them as "a central place to showcase their latest articles, videos and social posts." A beta Insights section provides performance data including impressions, clicks, and audience reach by country, powered by Search Console.

The release sits inside a longer sequence of changes to how Google surfaces and distributes publisher content through AI-powered feeds rather than traditional search results. For publishers managing both a Search presence and a Discover strategy, Search profiles add a formal identity layer to that relationship.

Also noted

  • June 4 Google announced the Meitner Energy Center in Texas, a data center built alongside more than 1 gigawatt of wind, solar, and battery storage on a single site, with air cooling designed to limit water use - part of a broader $40 billion Texas infrastructure commitment through 2027. PPC Land
  • June 5 Digiday reported that agencies are increasingly betting on entertainment content - television formats, branded programming, and creator partnerships - as a hedge against the continued decline of traditional display and social advertising effectiveness. Digiday
  • June 4 EDO launched Ad EnGage Optimize, a vertical AI tool for CTV that applies in-flight adjustments to frequency caps, creative rotation, and audience targeting, claiming to reclaim up to 24 percent of ad budgets otherwise lost to waste. PPC Land
  • June 4 Apple's App Store reached $1.4 trillion in developer billings during 2025, with AI-featured apps growing four times faster than the rest of the platform and advertising revenue from the ecosystem reaching $151 billion. PPC Land
  • June 4 Ray Dalio told a Bloomberg summit that US government borrowing, running approximately $2 trillion above receipts, has passed a structural inflection point - macroeconomic context that sits behind every infrastructure commitment discussed this week. The Spend

Timeline

  • June 6, 2026 - OpenAI activates ChatGPT advertising in the UK, with Japan, South Korea, Brazil, and Mexico to follow. PPC Land
  • June 5, 2026 - Digiday reports Liftoff Mobile raises $437 million in its Nasdaq IPO at $23 per share, the first major ad tech listing in more than 12 months. Digiday
  • June 5, 2026 - PPC Land publishes a synthesis of the week's infrastructure and commerce media announcements, tracing the chain from compute spending to advertising surfaces. PPC Land
  • June 5, 2026 - MediaPost reports Pinterest commits $4 billion to AWS through 2031 to power visual search AI, the largest infrastructure deal in the company's history. MediaPost
  • June 5, 2026 - Search Engine Roundtable confirms Google Search profiles are officially live, with Discover distribution tied to the follow mechanism. Search Engine Roundtable
  • June 4, 2026 - AdExchanger reports Roblox names SuperAwesome as exclusive ad partner for under-13 users, with direct-deal-only inventory and no programmatic access. AdExchanger
  • June 4, 2026 - DoorDash announces six simultaneous ad platform updates, including the Spotlight homepage format, Symbiosys offsite reach, and a LiveRamp clean room integration. PPC Land
  • June 4, 2026 - Google launches Search profiles in the US, giving publishers a claimable page that feeds into Discover distribution. PPC Land
  • June 4, 2026 - Pinterest commits $4 billion to AWS through 2031 to train visual AI models on Trainium and run inference on Graviton, covering 631 million monthly active users. PPC Land
  • June 4, 2026 - EDO launches Ad EnGage Optimize, a CTV campaign management tool claiming up to 24 percent budget recovery through in-flight frequency and creative adjustments. PPC Land