AI ad spend forecast questions ChatGPT and a NOK 20m GDPR fine land
EMARKETER puts ChatGPT US chatbot ad ceiling at $5bn by 2030, Norway fines Elkjop NOK 20m for loyalty consent failures, and Amazon deploys World Cup on Fire TV.
Two data points arriving within hours of each other on June 9 defined the news cycle for digital advertising. The first was a detailed forecast from EMARKETER placing a firm ceiling on what chatbot advertising can actually generate over the next four years. The second was a regulatory fine from Norway that laid out, in clinical procedural detail, exactly how a large retailer mishandled consent in a loyalty program that used Customer Match. Neither story travels well in isolation. Together, they sketch a map of where AI advertising money actually flows, and where the legal fault lines run beneath the data infrastructure holding it up.
The chatbot ad market is much smaller than OpenAI implies
On June 9, 2026, EMARKETER published a forecast projecting that total US advertising spending in and around artificial intelligence platforms will grow from $32.03 billion in 2026 to $68.25 billion by 2030. That is a doubling of the market in four years. The headline figure sounds generous to the AI advertising thesis. The internal breakdown is not.
According to EMARKETER, more than 80% of that projected spending in 2026 flows next to AI content, not inside chatbot interfaces. The largest category in the taxonomy is AI search-adjacent advertising, which covers traditional keyword-based search ads appearing alongside AI-generated summaries such as Google AI Overviews. Those ads do not require chatbots, conversational interfaces, or any fundamentally new advertising technology. They are the existing search auction, operating in the same page space as a new type of result. The EMARKETER framework also covers AI conversational search advertising, referring to ads appearing inside search-based chatbot experiences such as Google AI Mode. The smallest segment is AI chatbot advertising, which covers standalone large language model platforms including ChatGPT, Microsoft Copilot, and Gemini.
The report, titled “US AI Advertising Forecast 2026,” was authored by Nate Elliott, EMARKETER’s Principal Analyst for AI in Marketing and Commerce, who joined the firm in August 2025 after nearly nine years running his own research consultancy, Nineteen Insights. Elliott’s projections for the standalone chatbot segment are blunt. According to EMARKETER, ChatGPT and its direct chatbot competitors will generate less than $1 billion in US advertising revenue in 2026. By 2030, that number reaches just over $5 billion. OpenAI has described a target of $100 billion in global ad revenue by 2030. Even granting significant international expansion beyond the US market, the EMARKETER numbers suggest the structural ceiling is far lower than that figure.
Three specific constraints explain the gap. First, chatbots cannot accommodate the advertising density of traditional search engines. According to Elliott’s analysis, to reach even the chatbot advertising numbers EMARKETER projects by 2030, platforms would already need to insert paid placements into nearly every commercial response. That level of ad load would conflict with the conversational experience that makes chatbots useful, creating a structural ceiling that search engines, with their more transactional user intent, do not face in the same way.
Second, chatbot CPMs are falling. According to EMARKETER, ChatGPT launched its advertising pilot at a $60 CPM. The forecast projects that price will decline to roughly $15 by 2030, approximately one-quarter of the launch rate. Reports from April 2026 documented CPMs dropping to $25 within weeks of the US pilot launch in February, suggesting the compression is already well under way and tracking ahead of the forecast curve.
Third, and most structurally important: consumers do not spend most of their AI time inside chatbots. The volume of interaction required to build a mass advertising market remains dispersed across search interfaces, embedded features, and AI-assisted applications rather than concentrated in standalone conversation windows. According to EMARKETER, it is this distribution of consumer attention, not the technical limitations of ad delivery, that sets the tightest constraint on chatbot ad revenue.
On June 9, Elliott posted the core finding on LinkedIn. The chatbot ad market, he wrote, is roughly one-tenth the size that OpenAI has projected for the US alone. That framing arrived on the same day that ChatGPT expanded its advertising pilot beyond the US to the United Kingdom, activating ads for free and Go-tier subscribers while keeping Plus, Pro, and Enterprise users entirely ad-free. OpenAI’s geographic expansion and EMARKETER’s market sizing do not directly contradict each other. But they suggest sharply different theories of where advertising value will ultimately concentrate.
Earlier Barclays analysis published June 7 placed ChatGPT’s global ad revenue potential at $102 billion by 2030. For context, Google Search and YouTube ads combined generated $261 billion in revenue in 2025. OpenAI has neither the inventory scale of a search engine with decades of auction infrastructure nor the video advertising reach of YouTube. The EMARKETER analysis makes those gaps concrete in numerical terms. The broader AI ad spend total of $68.25 billion by 2030 remains significant. What the breakdown signals is that the money concentrates in the hands of companies that already operate large AI-adjacent content surfaces, rather than in newer entrants building chat products.
Elkjop, Customer Match, and four GDPR violations
On June 9, 2026, Norway’s data protection authority Datatilsynet published its decision fining Elkjop Nordic AS and Elkjop Norge AS a combined NOK 20 million for unlawful personal data processing in the company’s customer loyalty program. The decision, dated June 1, followed an on-site inspection that began in June 2022. The investigation ran for nearly four years.
The case matters beyond its fine size. Elkjop is a major Nordic electronics retailer with more than six million loyalty club members across the region. The loyalty program offered members general discounts, personalised offers, newsletters, and exclusive access. Joining required accepting all associated data processing at once. No granular options. No separate consent for marketing communications versus profiling versus analytics. According to Datatilsynet, Elkjop’s own compliance team described the approach as “all or nothing” internally, a phrase documented in the ruling.
The authority found the consent invalid on three cumulative grounds. First, it was not specific: separate processing purposes were bundled into a single acceptance, with no mechanism for customers to consent to, say, newsletters while declining profiling. Second, it was not freely given: refusing all processing meant exclusion from the membership program entirely, creating an imbalance of power that GDPR consent requirements are specifically designed to prevent. Third, it was not informed: checkout materials emphasised discounts and membership benefits while omitting clear explanation of the profiling and analytics activities the membership triggered.
According to Datatilsynet, the quality of disclosure at store checkouts also varied depending on which individual employee was serving the customer. That structural inconsistency meant that two customers joining the loyalty club at the same store on the same day might receive meaningfully different information about what they were consenting to. The authority noted that Elkjop had internally identified this risk as early as February 2022, before the inspection began, and had continued operating the system without structural changes.
The second violation concerned Customer Match. According to the ruling, Elkjop was in a test period for the tool at the time of the June 2022 inspection, with plans to begin uploading loyalty club contact data to advertising platforms in mid-2022. The stated purpose was to improve targeting efficiency, reach specific customer segments, and improve media investment returns. Datatilsynet found this constituted a new processing purpose that required a compatibility assessment under Article 6(4) of the GDPR. Elkjop had not conducted any such assessment, on the grounds that it considered the original and new purposes identical. The authority rejected that position.
The legal mechanism is worth holding for a moment, because Customer Match is not a niche tool. It is among the most widely used audience-matching products in digital advertising, connecting first-party data with platform user profiles across Google Ads, Meta, and other major systems to allow retargeting, exclusion lists, and lookalike modelling. According to Datatilsynet, uploading data from a consent-based loyalty program to a Customer Match tool, without a documented purpose compatibility assessment, is unlawful under the GDPR’s purpose limitation principle in Article 5(1)(b). That reasoning applies to any brand or retailer running a structurally similar architecture.
The authority also rejected Elkjop’s secondary argument that legitimate interest under Article 6(1)(f) could serve as a fallback legal basis for Customer Match processing. According to Datatilsynet, the fairness principle in Article 5(1)(a) requires that processing not occur in ways that are unexpected or misleading from the data subject’s perspective. Customers who joined a loyalty club to receive discounts could not reasonably have anticipated that their contact details would later be uploaded to advertising platforms for audience matching purposes.
Google has been migrating Customer Match uploads to the Data Manager API during 2026, a technical change that streamlines upload compliance documentation. It does not, however, resolve the underlying consent and purpose questions the Elkjop ruling has now formalised. The technical path for uploading data has changed; the legal obligations around what data can be uploaded, and on what basis, remain unchanged.
The third violation concerned offline conversion tracking documentation. Elkjop’s records of data flows between in-store purchases and digital advertising platforms were insufficient under the GDPR’s accountability obligations. The fourth violation involved children’s personal data. The minimum club membership age was 15, but the company lacked age verification mechanisms, did not record customer age data, and sold gaming products that attracted younger audiences. According to Datatilsynet, children were included in the profiling not incidentally but structurally, as a predictable consequence of the product range and the age of the customer base.
A combination of factors kept the fine below the maximum available. Elkjop cooperated with the investigation and made changes to its consent design during the proceedings. The authority acknowledged those remediation steps. The NOK 20 million figure nonetheless represents one of the most technically detailed GDPR rulings to address Customer Match specifically. Its reasoning on purpose limitation, consent bundling, and audience matching tool use will carry weight in future cases.
The Elkjop decision connects directly to a separate development that PPC Land covered on June 8. A court in Dusseldorf has referred a question to the Court of Justice of the European Union asking whether GDPR Article 26(2) requires joint controller terms to be publicly accessible online. That question, if answered affirmatively, would compel ad tech platforms to publish the data-sharing agreements that currently govern how data moves between demand-side platforms, supply-side platforms, and data management systems. The Elkjop case addresses one side of the consent chain; the CJEU referral addresses the transparency of the infrastructure on the other side.
Amazon, Fire TV, and the World Cup distribution race
On June 9, 2026, Amazon activated a dedicated FIFA World Cup 2026 hub on Fire TV, integrating FOX One as the primary streaming destination for all 104 tournament matches in the United States. The hub is accessible through the navigation bar, the sports tab, or the platform’s home screen featured content. According to Amazon, the design eliminates the need to switch between individual applications to locate a specific match.
The 2026 tournament is substantially larger than any previous edition. The competition features 48 teams in 104 matches from June 11 through July 19, up from the 32-team, 64-match format of prior editions. Three host nations hold their own opening ceremonies across 16 cities in the United States, Canada, and Mexico. FOX One, the official English-language streaming service for the tournament in the US, carries all 104 live matches along with daily highlights and full replays. A subscription is required for the full broadcast. Two matches are available at no cost through Tubi: Mexico versus South Africa on June 12 at 3 p.m. ET, and the US Men’s National Team’s opening match against Paraguay on June 13 at 9 p.m. ET.
Tubi, owned by Fox Corporation and operating as a fully ad-supported platform, surpassed 100 million monthly active users in May 2025. Its model, where all content is free in exchange for advertising exposure, makes it the natural vehicle for extending World Cup reach to unsubscribed households. According to Amazon, the Fire TV hub also integrates Alexa+ voice navigation in nine countries, enabling users to access schedules, scores, and live matches through voice commands from the home screen.
Bauer Media Audio was announced as an official Fire TV Channels launch partner on June 9, adding KISS, Hits Radio, and Irish radio brands to the platform free of subscription requirements. The timing places a significant audio expansion alongside the World Cup launch, extending Fire TV’s proposition into live radio at the same moment tournament viewing peaks household usage.
The World Cup distribution story has been building across multiple surfaces for weeks. FOX Sports and ReachTV will carry all 104 matches across more than 80 US airports, reaching 51 million monthly travellers from June 11 through July 19. Google activated live scores, agentic ticket booking, Street View stadium imagery, and Gemini match briefings across Search, Maps, Waze, and the Gemini app on June 8. The tournament has become a coordinating event across media, technology, and out-of-home advertising industries simultaneously, with each new platform announcement illustrating how connected television, AI search, and digital out-of-home are converging around the same audience.
Semrush maps how LinkedIn earns AI citations
On June 9, 2026, Semrush published a detailed analysis explaining how B2B brands can improve their visibility in AI search tools by optimizing LinkedIn content. The guide draws on 89,000 LinkedIn URLs cited by ChatGPT Search, Google AI Mode, and Perplexity, collected from 325,000 unique prompts submitted between January and February 2026. The research was authored by Margarita Loktionova, Product and Content Marketing Lead at Semrush, and builds on an earlier dataset published in March 2026 that identified LinkedIn as the second most-cited domain in AI search responses, appearing in 11% of AI-generated answers on average across the three platforms. Only Reddit was cited more frequently.
According to Semrush, citation behavior varies significantly by platform, and that variance has direct implications for content strategy. Perplexity cites LinkedIn Company Pages 59% of the time. ChatGPT Search and Google AI Mode, by contrast, cite individual creator profiles 59% of the time. A company that publishes only on its corporate page will be well-positioned for Perplexity but largely absent from the citation pools of the other two major platforms. The playbook argues for a dual-track approach: corporate content to capture Perplexity’s preference, employee content to address the preference of ChatGPT Search and Google AI Mode.
Article format carries significant weight. According to Semrush, LinkedIn articles account for between 50% and 66% of citations across all three platforms. Shorter posts account for 15% to 28%, depending on the model. Neither format dominates entirely. Articles tend to be surfaced for in-depth answers; shorter posts appear for more discrete, specific questions. A content strategy built exclusively on long-form articles or exclusively on short posts leaves a substantial share of potential citations uncaptured.
The Semrush methodology recommends starting from customer-facing sources: sales transcripts, demo recordings, support tickets, and win-loss notes. These inputs produce content anchored to demonstrated demand rather than assumptions about what might rank. The approach is consistent with earlier independent research. According to an analysis by Peec AI published June 6, which examined 5 million ChatGPT fanout queries, ChatGPT systematically injects terms such as “best,” “reviews,” and the current year into the hidden searches it runs before generating a response. That pattern rewards content built around comparison, evaluation, and specificity rather than broad brand positioning.
The timing of the Semrush analysis connects directly to the EMARKETER forecast. If more than 80% of AI advertising spend flows next to AI-generated content rather than inside chatbots, then the question of which brands appear inside those AI responses, and how frequently, becomes a meaningful commercial variable. LinkedIn’s position as the second most-cited domain in AI search places it inside that dynamic. Brand visibility in AI citations is not yet a line item in media plans, but the data on citation frequency, content format, and platform-specific behavior is now specific enough to act on.
Also noted
June 9, 2026: Google Analytics added a direct integration with Google Business Profiles on June 8, importing seven local interaction metrics, including calls, bookings, direction requests, and website clicks, into Analytics reporting alongside web and app data. Historical data is limited to a six-month rolling window from the date of linking. ppc.land
June 9, 2026: Google upgraded NotebookLM on June 8 with Gemini 3.5, a secure cloud computer, more than 100 software skills, and 11 new downloadable output formats including Word, PDF, and audio, available to paid tiers immediately. ppc.land
June 8, 2026: Apple announced Siri AI, built on Google Gemini models, at WWDC 2026, alongside new App Store Creative Assets formats, updated ad coordination tools, and an EU rollout delayed by regulatory review, reshaping app advertising mechanics. ppc.land
June 8, 2026: A German court in Dusseldorf referred a question to the Court of Justice of the European Union asking whether GDPR Article 26(2) requires joint controller terms to be publicly accessible online, a ruling that could force ad tech platforms to publish data-sharing arrangements currently kept private. ppc.land
June 8, 2026: Tubefilter, Comscore, Whalar Group, and Gospel Stats launched The Creators List, a curated directory of more than 200 creators attending Cannes Lions this month, built for direct brand-creator introductions ahead of the festival. ppc.land