China's MIIT to Cultivate 2,000 AI Service Providers; ChatGPT Ad Revenue Run Rate Surpasses $1B in 200 Days

China's MIIT Launches AI Service Provider Cultivation Initiative

On August 31, China's Ministry of Industry and Information Technology (MIIT) issued a notice officially launching a special initiative to cultivate AI application service providers. Widely seen as a pivotal government-level push to accelerate AI adoption, this move signals a strategic shift in China's AI industry from "technology R&D" toward "scaled application deployment."

According to the notice, the initiative follows four guiding principles: expanding quantity while improving quality, tiered cultivation, innovation-driven development, and application orientation. The core targets are clearly defined: establish a national service provider resource pool, grow it to over 2,000 providers by end of 2026, and reach at least 3,000 by end of 2027. These quantified goals reflect a sense of urgency at the policy level around building a robust AI application ecosystem.

Notably, the notice outlines four key tasks: building the resource pool, raising supply-side capabilities, driving scaled application, and strengthening support infrastructure. One particularly practical detail encourages service providers to build "frontline deployment engineer teams" embedded at customer sites to address real enterprise needs directly. This reflects policymakers' recognition that the biggest bottleneck in AI adoption is often not the model itself, but translating technology into usable, industry-ready solutions.

The notice also calls for leveraging policy instruments such as "compute vouchers" to lower infrastructure costs. For small and mid-sized AI service providers, compute costs have long been a major barrier to commercialization. Policy subsidies are expected to reduce the industry's overall entry threshold and accelerate the formation of a tiered service provider ecosystem.

ChatGPT Ad Business Hits $1B ARR in 200 Days

Across the Pacific, OpenAI's commercialization momentum is equally impressive. According to disclosures, ChatGPT Ads has reached an annualized revenue run rate (ARR) of over $1 billion — achieved in roughly 200 days since the product launched in test mode in the United States in February of this year.

ChatGPT advertising now covers more than 40 countries and regions, with its self-serve ad platform this week expanding access to India, Europe, the Middle East, and North Africa. Underpinning this ad business is ChatGPT's user base of over 1 billion weekly active users — a massive traffic pool no advertiser can afford to ignore.

OpenAI's ad strategy is notably restrained. Ads are shown primarily to free-tier and lower-cost subscription users, and ad content is physically separated from AI-generated responses, preserving the integrity of generated output. Crucially, OpenAI has committed that advertisers cannot access users' private conversations — a key trust-building feature in an era where AI-related privacy concerns run high.

Looking ahead, OpenAI's ambitions are substantial: the company targets $2.5 billion in full-year ad revenue for 2026, with a long-term goal of $100 billion by 2030. This positions advertising as OpenAI's third major revenue pillar alongside subscriptions and API access, and charts a monetization path for the broader AI industry beyond pure subscription models.

Chemicals and Entertainment: Vertical AI Adoption Accelerates

Driven by both policy tailwinds and capital, AI applications in vertical industries are rapidly maturing.

Chemical Industry LLM 3.0 Pro Launched

Also on August 31, the Chemical Industry Large Model 3.0 Pro was officially released, jointly developed by the Dalian Institute of Chemical Physics (DICP) under the Chinese Academy of Sciences, iFLYTEK, Alibaba Cloud, and other partners. As the first iteratively upgraded large model in China's chemical industry, it marks a leap from "professional knowledge Q&A" to an "intelligent chemical task execution system."

The model is built on a four-layer technical architecture — large model + intelligent agents + professional skills & tools + application scenarios — integrating over 400 chemical agents and tools to form a complete loop of "understanding → planning → execution → verification." Performance metrics include an 81.96% accuracy rate for text-based Q&A and 80.75% for multimodal Q&A, representing overall score improvements of 20.2% and 31.4% over the previous generation, respectively. Over 300 enterprises, universities, and research institutes have registered to use the platform, with cumulative API calls surpassing 14 million — a testament to solid real-world adoption.

First AI Long-Form Drama Premieres in Prime Time on Satellite TV

Also on the evening of August 31, Post-Production by Mail (《后期邮寄》), produced by Mango TV and billed as China's first AI-generated long-form drama, officially premiered during Hunan Satellite TV's prime-time slot. The series features no human actors — all visuals and character performances are generated by AIGC — and is the first production to adopt a "simultaneous production, review, and broadcast" workflow following the release of the broadcasting authority's "21 Provisions on Generative AI Content."

Most disruptive is its cost structure: the production cycle was just four months, at roughly one-tenth the cost of a live-action drama. The news sent Mango SuperMedia shares to back-to-back limit-up days, adding nearly 11.7 billion yuan in market cap over two sessions and lifting the broader AI film and television sector. Industry observers widely view this as AIGC content formally crossing from the experimental stage into the mainstream television content industry.

Amazon AWS Doubles Down on Government AI

At the infrastructure layer, Amazon AWS announced an expansion of AI model availability within its GovCloud (US) regions. Major global frontier models from OpenAI, Anthropic, Meta, xAI, and NVIDIA are now collectively accessible via the Amazon Bedrock platform, alongside Amazon's proprietary Nova model series — all running within isolated cloud environments that meet government compliance requirements.

This strategic positioning is backed by strong commercial data: AWS reported Q2 FY2026 revenue of $42.2 billion, up 36.7% year-over-year — its fastest growth rate in 18 quarters. Both its AI business and proprietary chip business have surpassed $25 billion in ARR. To sustain expansion, Amazon has committed up to $50 billion in capital expenditure for 2026 to add approximately 1.3 gigawatts of AI and high-performance computing capacity across multiple secure cloud regions in the United States.

The government sector has always been a fiercely contested market for cloud providers — high compliance barriers, strong customer stickiness, and large contract sizes. By bringing mainstream large models into compliant cloud environments, AWS is effectively lowering the barrier to AI adoption for government and sensitive-industry customers, securing an early lead in this high-value segment.

Conclusion: From Technology Narrative to Deployment Race

Taken together, these developments paint a consistent picture of where the AI industry is headed: the competitive focus is shifting from "whose model is better" to "who can deploy AI faster, cheaper, and more compliantly."

MIIT's service provider cultivation initiative targets the application ecosystem; OpenAI's ad monetization explores new business model innovation; the chemical industry LLM and China's first AI prime-time drama validate vertical deployment capabilities; and AWS is fortifying its moat at the infrastructure layer. Whether from a policy, capital, or technology perspective, all signals point in the same direction — the decisive phase of scaled AI application deployment has arrived.