Aakash Gupta, founder of the Product Growth newsletter and podcast, has highlighted the growing influence of Chinese open-source models in the startup ecosystem. According to Gupta, these models are gaining traction due to their significantly lower costs, which make fine-tuning and specialization a default strategy for startups. This statement was made on X.
"The real constraint in AI was never capability," said Gupta, Author & Host. "Chinese models are matching GPT-4 on coding benchmarks while costing 2% as much. The constraint was always burn rate, and China solved it first by optimizing for efficiency instead of chasing AGI. China optimized for distribution and adoption. Making AI cheap enough to become infrastructure."
At Fortune’s Global Forum, Andreessen Horowitz general partner Anjney Midha noted that open-source artificial intelligence (AI) is currently dominated by China. Midha explained that investors are observing founders prioritize affordability and rapid iteration over marginal quality improvements. This preference is leading early-stage teams to increasingly adopt lower-cost Chinese models, particularly when expanding AI into new areas such as cryptocurrency and Web3 technologies.
Reuters reported that DeepSeek disclosed a $294,000 training bill for its R1 reasoning model using 512 Nvidia H800 chips. This cost is significantly lower than the "much more than $100 million" figure cited by Sam Altman for advanced U.S. models. While not all comparisons may be directly equivalent, this figure highlights the substantial cost difference that allows startups to achieve per-token pricing as low as pennies.
Coinbase suggests that AI innovation is driving advancements in cryptocurrency through "onchain AI," where agents can autonomously manage wallets and programmable payments. By facilitating micropayments and coordinating services on blockchains, these agents enable business models like metered APIs and usage-based royalties that were previously unfeasible with traditional payment systems.
Similarly, Binance is integrating AI with "Bibi," an assistant designed to condense crypto whitepapers into brief summaries within the exchange's app. By reducing research time and cognitive effort required for due diligence, Bibi helps users quickly evaluate tokens and experiment more freely. This development illustrates how AI enhancements are accelerating Web3 adoption by linking cryptocurrency with AI innovation.
Gupta operates "Product Growth," an AI product management newsletter and podcast with a substantial following. With over 15 years of experience in product roles at companies such as Epic Games, Affirm, and thredUP, Gupta now focuses on practical growth strategies for builders. His analysis emphasizes cost-efficiency, experimentation, and financial planning—key considerations for both AI and crypto entrepreneurs.