3 days ago
India a major player in AI adoption, but China surging on open-source AI: Report
Mary Meeker, often called the 'Queen of the Internet' for her influential tech trend reports in the 1990s and 2000s, is back with a deep dive into artificial intelligence. Her latest report — Trends: Artificial Intelligence — spans 340 pages and is packed with charts, predictions, and sharp insights. If one word stands out, it's 'unprecedented.' Meeker uses it 51 times, and not without reason. 'The pace and scope of change related to the artificial intelligence technology evolution is indeed unprecedented,' she writes. The report shows how AI adoption is moving faster than any technology before it, with billions of users, soaring capital spending, and increasing global impact. From how we search for answers to how we work, code, create, and communicate, AI is now part of everyday of the clearest signs of this shift is the explosive growth of ChatGPT. While apps like Instagram and YouTube took two to four years to reach 100 million users, ChatGPT did it in under three months. As of April 2025, it had 800 million weekly users and processed more than 365 billion searches annually — more than Google managed in its first leads in adoptionIndia is playing a crucial role in this growth. According to Meeker's report, the country contributes the highest share (13.5 per cent) of ChatGPT's mobile app users, ahead of the US (8.9 per cent) and Germany (3 per cent). It's also the third-largest user base for DeepSeek, a Chinese AI platform, accounting for 6.9 per cent of its users. That's significant considering ChatGPT is banned in both China and Russia — the top two markets for
India's digital population and mobile-first internet habits are helping it emerge as a major AI consumer market, and one that global platforms can't afford to surges on open-source AIThe report also outlines a growing split between closed AI models and open-source systems. Closed models like GPT-4 and Claude dominate in performance and are popular among enterprises due to their ease of use and capabilities. But they're also more opaque, tightly controlled, and require massive capital to build and the other hand, open-source models such as Meta's Llama and Mistral's Mixtral are lowering the barrier to entry. Developers, startups, and even governments can download and fine-tune them without billion-dollar budgets. Meeker notes how open models are powering local language tools, community innovation, and sovereign AI China is leading the open-source race. In just the second quarter of 2025, it released three major models — DeepSeek-R1, Alibaba's Qwen-32B, and Baidu's Ernie 4.5.'We're watching two philosophies unfold in parallel — freedom vs control, speed vs safety, openness vs optimisation,' Meeker writes. 'Each is shaping not just how AI works, but who gets to wield it.'What lies aheadDespite the buzz, Meeker strikes a note of caution. While AI platforms have amassed huge user bases, most still earn relatively low revenue per user, a median of just $23 (around Rs 2,000). Investors are betting big, but the business models are still training large models has become increasingly expensive, sometimes costing up to $1 billion, though the cost of using them is dropping fast. Chips from Nvidia, Google, and Amazon are getting more energy-efficient and powerful, fuelling this it all is a race for infrastructure — chips, GPUs, data centres — that Meeker compares to the space race. With both China and the US leading in development, she says the mix of technological progress and geopolitical tension makes this era one of both opportunity and uncertainty.
