$250 million. $4 billion. $85 billion.
That’s how much companies like xAI, Google, and OpenAI are paying to train and run AI models like Grok, Gemini, and ChatGPT, respectively – and it’s why Elon Musk is adding advertisements to X’s AI chatbot, Grok.
When Elon Musk joined his AI startup with his social network early this year, I began to question if X’s advertising business was still the primary focus. When Chief Executive Officer Linda Yaccarino, a long-time advertising professional, left the firm a few months later, I was confident that X’s ad revenue would decline. After several years of effort, it simply didn’t appear large or profitable enough to capture Musk’s attention.
So I was caught off guard earlier this August when Musk unexpectedly sponsored an hour-long AMA on X to discuss advertising. Given his concentration on Tesla, SpaceX, and federal government cost-cutting at DOGE, Musk had not publicly moved his attention to X’s advertising business in quite some time. It was purposeful. In an interview, Monique Pintarelli, X’s head of ad sales for the Americas, explained that the business intended to “address the whispers from the press and the industry that in some way X is deprioritising advertising.” “That’s actually completely false.”
Musk spent a large portion of his AMA discussing how the firm intended to leverage its merger with his artificial intelligence company, xAI, to boost its advertising offerings. He intends to employ Grok, xAI’s chatbot, to construct user targeting profiles so that the firm can better understand what items people are interested in. He also wants Grok to assist in analysing ads before they run to ensure they meet “some reasonable aesthetic standard” and do not appear to be an “eyesore.” The corporation stated that ads with a better appearance will receive higher placement in X’s advertising auction.
According to Musk, Grok will leverage vector embeddings and cosine similarity (a concept gaining popularity in OmniSEO) to send adverts to the appropriate user at the right time. As these ads remain in the ad network, their user match rate will improve.
Eventually, Pintarelli stated, X hopes to give marketers an all-in-one solution in which they can simply show up with a picture of their product and X’s AI tools would handle the rest, including developing the ad copy and targeting the product to the proper individuals — all using AI. Others in the business, notably Meta Platforms Inc., are pursuing a similar strategy.
Musk stated that adverts may surface when a user asks Grok an inquiry about an issue, and the chatbot could recommend items or services that directly address the problem. According to him, this method would make advertisements more valuable and relevant rather than just promotional. He stated that the strategy is also related to addressing the high expenditures of running Grok. The AI system relies on powerful and pricey chips known as graphics processing units (GPUs) to function.
It’s reasonable to wonder if AI-enhanced marketing will genuinely make a difference for a company dealing with much larger challenges. It doesn’t matter how sophisticated the advertising technology is if firms are concerned about being associated with Musk or X. Grok has received some unfavourable publicity. In recent months, xAI had to apologise after an upgrade that prompted the chatbot to publish objectionable information on X. These featured antisemitic jokes and statements supporting Adolf Hitler, which were heavily criticised. Musk did not dismiss these concerns, but emphasised that Grok remained superior in intelligence and accuracy when compared to other AI chatbots.
It is unclear how much advertising on Grok will cost, whether based on cost-per-click (CPC), cost per mille (CPM), cost per acquisition (CPA), or another bidding strategy. AI search advertising is anticipated to reach $25 billion by 2029 as more customers utilise AI chatbots such as Grok, ChatGPT, and Perplexity for search. Conversational advertising on Grok may result in cheaper costs for marketers due to increased relevance and niche demographics. In comparison, Microsoft Copilot sponsors have already quadrupled their return on advertising expenditure.
Advertising in AI experiences is not new to Grok. Microsoft Copilot and Perplexity have already launched sponsored advertisements (but only to chosen advertisers), while ChatGPT ads are expected to appear in 2026. The outcomes of these conversational advertising tests, however, are encouraging. According to Microsoft, Copilot search advertising is 25% more successful than standard search ads. Notably, more than 40% of consumers had a favourable experience with these adverts, demonstrating that marketers can build appealing offers that AI platforms can send at the appropriate moment.
The logic underlying this desperate action is delightfully plain. X’s UK operations witnessed a 66.3% decline in sales in the year after Musk’s acquisition, from £205.3 million in 2022 to £69.1 million in 2023. Meanwhile, ad income fell by over half (46.4%) between 2022 and 2023, from $4.5 billion to $2.2 billion per year, with WARC predicting a total loss of $5.9 billion in ad revenue since Musk’s acquisition.
But here’s the kicker: only 4% of marketers thought advertising on X was brand-safe, and a quarter (26%) want to cut expenditure by 2025. So, Musk’s amazing solution? Double down on advertising by putting it in the one place on his platform that people may truly trust—his AI chatbot. Meanwhile, other social media sites showed significant development within the same time period. Instagram rose 24.9%, Snapchat by 13.8%, and Pinterest by 18.1%.
The timing is very rich. While OpenAI’s Sam Altman describes advertisements as a “last resort” and adds, “I think that ads-plus-AI is sort of uniquely unsettling,” Musk is speeding towards the nightmarish future that other AI pioneers are attempting to avoid. OpenAI CEO Sam Altman has been sceptical about advertisements, describing them as loud distractions that may influence AI results. Several advertising executives told the Financial Times that they want AI startups to offer ad spots in their chatbots. At the same time, groups like OpenAI have explicitly denied any plans to do so, citing concerns about a customer response.
The basic issue with advertising in AI chatbots is not only aesthetic; it is philosophical. When people ask Grok a question, they want an objective response based on data and reasoning. When you include paid placements, every response becomes questionable. Did Grok promote that restaurant because it is genuinely impressive or because they paid for the privilege?
This is especially problematic for a platform that already has trust concerns. According to Kantar research, X’s trust score has plummeted from 28% in 2021 to 16% in 2023. Adding sponsored material to AI answers is akin to attempting to repair trust by lying more effectively. Musk mentioned a “solution” in which “advertising the specific solution would be ideal” while users are attempting to solve problems, which sounds eerily similar to turning Grok into a glorified shopping assistant. Imagine asking about the best approach to control anxiety and receiving a sponsored response from some dodgy supplement firm.
The humorously predictable aspect of Musk’s statement is his attempt to portray it as innovative. He discussed intentions to employ Grok technology to “assess how aesthetically pleasing adverts are and prioritise those deemed more so.” Because what consumers truly want from their AI is more appealing advertisements that disrupt their hunt for information. Meanwhile, on the same call when Musk unveiled his advertising plans, some advertisers offered him “glowing reviews” of X’s changes. In contrast, others refused to participate, claiming that the platform’s material was still too toxic. As one media buyer at a major holding firm told the Financial Times, “I’m not wasting my time.”
This was always bound to happen. When you acquire a platform for $44 billion and then methodically destroy its income sources while incurring $1.2 billion in yearly debt service payments, you finally run out of options for avoiding the apparent answer. X currently has around 1.3 million paying members (0.26% of its entire user base), generating approximately $176.4 million in annual subscription revenue—a rounding error when compared to its commitments.
The actual problem is not that Musk is bringing advertisements to Grok. It’s that he’s doing precisely what every desperate platform owner does when they’ve run out of other options: sacrificing user experience for short-term profit. It’s the same approach that has ruined numerous platforms in the past, repackaged with Silicon Valley buzzwords like “AI-powered targeting” and “aesthetic optimisation.”
The only thing that surprises me is how surprised everyone is. This was always the final goal for a platform run by someone who profoundly misunderstands the link between content moderation, advertiser confidence, and long-term profitability. We’re witnessing a masterpiece in how to transform a $44 billion acquisition into an expensive lesson in the significance of not alienating your core income base while claiming to innovate.
