Mary Meeker’s ‘AI Trends’ report: What the “Queen of the Internet” says about ChatGPT, China’s AI race, and jobs in 2030

Mary Meeker - AI Trends report

Investor and analyst Mary Meeker — also known as the “Queen of the Internet” for her era-defining Internet Trends reports — has dropped a 340-page deep dive into the state of artificial intelligence. Titled Trends – Artificial Intelligence, the report outlines how AI is rapidly reshaping global behaviour, industries, and economics. From the explosive rise of ChatGPT to China’s sovereign AI push and the future of work, here are 10 key takeaways from one of the most comprehensive AI reports published so far in 2025.

AI is growing faster than the internet ever did

We’ve heard this before, but Meeker shows it clearly: the speed at which AI tools are being adopted globally makes the early internet look slow. Developers are shipping faster, users are adopting faster, and platforms are learning in real time. Everything is compounding — smarter tools are helping build even smarter tools.

ChatGPT is the fastest-growing product ever

Not just in AI — across all consumer tech. It took ChatGPT two months to hit 100 million users. That’s faster than TikTok, Instagram, Netflix, or the iPhone. By April 2025, it had 800 million weekly users and now handles over 365 billion searches annually. Meeker compares it to Google — except it did in two years what took Google more than a decade.

Freemium is the go-to model, but infra costs are biting

Most major AI apps are free to try, but charge for power use, better results, or API access. Think ChatGPT Plus, Midjourney’s subscription tiers, or Claude’s API. This helps scale quickly, but Meeker flags a concern: compute costs are massive. Training and inference (especially with GPUs) are expensive. Monetisation has to catch up fast, or things break.

China’s not playing catch-up anymore

The report states that Chinese LLMs like Qwen2.5 (Alibaba) and CodeFuse (ByteDance) are matching or beating Western models in benchmarks. The local user base is massive, and the government is backing sovereign AI hard. China doesn’t want to rely on US infrastructure, and this new AI race is looking more like a tech Cold War.

AI-native products are replacing AI-as-a-feature tools

We’re not just adding AI to existing apps anymore. Entire tools are being built around AI from day one. Think Notion AI, GitHub Copilot, Perplexity, and AI-first design platforms. Meeker calls this the “platform shift.” It’s happening now, not 10 years from now.

Jobs aren’t disappearing — they’re changing, and it is happening fast

The panic around “AI is taking our jobs” is real, but the report takes a more nuanced view. Roles are evolving. AI is becoming a co-pilot for writers, coders, designers, and analysts. We’re already seeing a 448% rise in AI-related job postings since 2018 — even as other job listings are dropping. By 2030, you won’t need to be an engineer, but you’ll need to work with one (or one built into your tools).

Enterprises and governments are now fully in

AI is no longer a pilot project for large companies — it’s part of daily operations. Meeker cites examples from JPMorgan, Microsoft, Yum! Brands, even the US FDA. Governments are also pushing AI strategies, especially around health, defence, and regulation. Everyone wants in — and no one wants to fall behind.

AI is not just on your screen — it’s becoming tangible

Self-driving cars in San Francisco, smart robotics in factories, kitchen automation, health diagnostics — AI isn’t just typing prompts into chatbots. It’s powering real-world automation. The jump from “digital assistant” to “physical AI agent” is already here, and Meeker argues it’s just getting started.

The AI infrastructure race is the new oil race

Access to compute — GPUs, chips, data centres — is now a strategic advantage. Meeker compares it to the Cold War space race. Countries and big tech firms are pouring billions into infrastructure. Whoever controls computing controls the future of AI. This also explains why Nvidia’s valuation keeps soaring.

The ethics conversation is way behind

With all this growth, the report doesn’t ignore risk. Meeker points out the usual problems: hallucinations, bias, misinformation, and the open-source vs closed-source debate. Regulation is still catching up. She calls for transparency, thoughtful leadership, and smarter frameworks to keep this thing in check before it breaks trust at scale.