AI Race Heats Up: Mary Meeker Warns US Firms Face Pressure from Cheaper Rivals Like DeepSeek

Artificial Intelligence is evolving at an unprecedented pace, and star investor Mary Meeker, known for her keen insights into tech trends, is sounding the alarm. Her latest report highlights the growing challenges faced by US AI companies, particularly those developing large language models (LLMs).

Meeker, an early backer of tech giants like Meta and Spotify, believes that the next wave of AI innovation will create numerous companies worth $10 trillion. However, she cautions that these companies won't all be based in North America. Competition is intensifying, especially from cheaper rivals like China’s DeepSeek, potentially undercutting US firms.

According to Meeker, the cost of training advanced models like OpenAI's GPT series and Google's Gemini is rising significantly. Meanwhile, increasing competition, both domestic and international, puts pressure on pricing. She noted that smaller, cheaper models are emerging for custom-use cases, questioning the sustainability of a one-size-fits-all LLM approach.

Meeker's analysis underlines that general-purpose LLMs have economics that resemble commodity businesses with venture-scale burn. The explosion of generative AI tools since the launch of ChatGPT has spurred massive investment in data centers and AI infrastructure. The six largest US tech companies increased capital expenditure by over 60% year-over-year to over $200 billion in 2024. OpenAI, xAI, and Anthropic now have a combined valuation of around $400 billion.

Chip and algorithm upgrades have slashed the costs of running models, enabling rivals like DeepSeek to offer cheaper, more efficient AI. While training the most advanced models has become exceedingly expensive (estimated costs have increased 2,400-fold in eight years), Meeker warns that OpenAI's valuation-to-revenue multiple looks expensive.

The rush to $12 billion in annualized revenue for OpenAI, Anthropic, and xAI required a combined $95 billion, reflecting the high cost of entering and succeeding in the AI space. Competition favors consumers as they enjoy better features at lower prices, it also means that start-ups must have large cash reserves and a lot of patience.

Meeker advises investors to only invest what they're willing to lose and suggests a portfolio approach to AI. The rules that well while we are in euphoria mode is that it's a risk now to put all your eggs in one basket, because everything is up and to the right - until it isn't.

The AI race is far from over, and the ultimate winners remain to be seen. Will US companies maintain their lead, or will cheaper rivals disrupt the market? Is this new wave the start to something amazing?

What are your thoughts on the future of AI? Share your opinions in the comments below!

• DeepSeek
• Mary Meeker
• OpenAI