OpenAI ad business hits $1B run rate as ChatGPT gets EU designation

OpenAI ad business reaches $1 billion annualized run rate

OpenAI’s advertising business has hit an annualized revenue run rate of $1 billion, roughly 200 days after the company began testing ads in ChatGPT in the US in February. The milestone signals a diversified business model as OpenAI prepares for a major IPO and faces pressure to justify its $852 billion valuation. Ads are now available in more than 40 countries, and a self-serve option launches today in India, Europe, the Middle East, and North Africa.

Ads appear for Go subscribers and free users, supplementing existing revenue from corporate customers, subscriptions, and usage-based APIs. OpenAI says the ads are clearly labeled, are not meant to influence its answers, and that advertisers get no access to private conversations. The ad business is intended to complement, not replace, existing revenue streams.

OpenAI says its ChatGPT ad business hits a $1 billion annual run rate →

ChatGPT becomes first AI chatbot designated under EU Digital Services Act

The European Commission on Monday classified ChatGPT as a search engine and designated it a Very Large Online Search Engine under the Digital Services Act, making it the first AI chatbot designated under the regime. The same announcement designated Reddit and Roblox as Very Large Online Platforms. ChatGPT reported 159 million monthly active users in the EU over the six months to March 2026, well above the 45 million threshold; Reddit reported 57.2 million and Roblox roughly 48 million.

The designation brings OpenAI, Reddit, and Roblox into a compliance regime requiring annual systemic risk assessments covering illegal content, protection of minors, fundamental rights, electoral processes, and public security, along with independent audits and data access for regulators and vetted researchers. The companies have four months to comply, with a deadline at the end of November. Fines for serious violations can reach 6% of global annual turnover. Henna Virkkunen, the Commission’s executive vice-president for tech sovereignty, said the designations mean the companies “will now be held to a higher standard of scrutiny and accountability.” Roblox’s obligations are especially significant given its large child and teen audience, while OpenAI must assess risks around electoral processes and public security. First risk assessments are due in November.

ChatGPT becomes the first AI chatbot designated under the DSA →

OpenAI tests outcome-based pricing, charging only when AI completes tasks

OpenAI has begun giving some large customers the option to pay only when its AI completes a task, such as handling customer support requests, The Information reported, citing a person with direct knowledge of the arrangements. The option had not been public before, and an OpenAI spokesperson declined to comment. The move adopts outcome-based pricing, a model pushed hardest by startups like Sierra and Fin, which charge only for tasks the AI completes without human involvement. Coding assistant Cognition promises corporate customers credits of up to $10 million if its software fails to deliver results worth at least what they paid.

AI products are expensive to run and have not yet accelerated revenue growth at software companies, according to The Information. Customer IT budgets are strained by competing tools such as Anthropic’s Claude, which recently pushed further toward usage-based billing. Salesforce lets companies choose how they pay for its AI product, Agentforce, including contracts tied to how much the AI lifts revenue or cuts costs. Salesforce CEO Marc Benioff described a model where a customer pays $2 because the AI brought in $20 or $40, rather than paying $2 simply for the software closing a set number of calls. Attribution becomes the sticking point as the model spreads; Stripe has published guidelines noting that a result could come from product changes, marketing campaigns, or seasonality. Adobe will also bill part of its newly bundled AI suite, CX Enterprise, by the value it creates, such as the number of ad campaigns completed.

OpenAI starts charging some customers only when its AI actually works →

China’s CXMT produces first HBM3E chips, closing AI memory gap

China’s top memory maker ChangXin Memory Technologies (CXMT) has begun producing HBM3E for the first time in small quantities, according to The Information, citing two insiders. HBM3E is a fast memory found in many current AI processors, consisting of stacked memory chips placed next to the processor, and is needed to train and run large AI models. This puts CXMT a generation behind Samsung, SK Hynix, and Micron, which are mass-producing HBM4.

Alibaba’s T-Head and Cambricon are testing the memory and plan to use it in products starting in 2027. That could lower a hurdle for China’s own AI chips, because US export rules limit purchases of advanced HBM chips. The Information also reported that CXMT is three to five years behind technically and struggles with low yields; SemiAnalysis estimated its yields at around 25 percent in June. CXMT raised $8.6 billion in its Shanghai IPO, but none of that is earmarked for HBM according to the prospectus.

China’s CXMT makes its first HBM3E chips, closing the AI memory gap →

Nvidia invests $3.5B in MediaTek to expand custom AI chip ecosystem

Nvidia is investing $3.5 billion into Taiwanese chipmaker MediaTek. As part of the deal, MediaTek will adopt Nvidia technology to help design custom chips for AI companies and hyperscalers that can be plugged directly into Nvidia-based data centers. The deal comes as AI companies and major cloud providers including Amazon, Google, Microsoft, OpenAI, and Anthropic invest in building their own chips to reduce reliance on Nvidia GPUs; deals like this allow Nvidia to cede ground on custom silicon while keeping its lead as the dominant data center scaffolding.

MediaTek will gain access to Nvidia’s NVLink Fusion ecosystem, including NVLink, the technology that lets chips communicate quickly with each other. Last week, Nvidia announced a similar partnership with Amazon Web Services without a direct investment: AWS will deploy an additional 2 million Nvidia GPUs and integrate NVLink Fusion. MediaTek has been building its custom data center ASIC operations and said in June it expects that business to generate $2 billion in revenue in 2026. The companies will also continue collaborating on DGX Spark, RTX Spark, and AI-powered software-defined vehicles. “AI is transforming every computing platform — from the world’s largest AI factories to the PC and the car,” said Jensen Huang, founder and CEO of Nvidia. “Together, we’re building platforms that bring NVIDIA accelerated computing to new markets and give customers the freedom to create differentiated AI systems at enormous scale.”

Nvidia’s $3.5B MediaTek bet reveals its plan for tackling Big Tech’s AI chip buildout →