The cheap replica market has the same principles. Doesn’t matter whether it is jeans or genes or ai. If you have no moat other than PR, sooner or later, you get called out. Of course it doesn’t make sense for India to join the replica market.
It makes sense to reinvent ai. Ground up. Configured for distributed low energy compute, virtualised affordable terminals and separation of data from models. This is not going to happen in a hurry. And govts can’t do breakthroughs. What they can do is NOT to give subsidies to bad actors which will make good actors unviable.
Ai subsidies if any should only be given to r&d startups using the aif framework. And most certainly not to bad actors like BigTech. There is no need to fall for narratives which essentially say that becoming a meal ticket for Nvidia will define your country’s progress. Best to give both Nvidia /cuda and Qualcomm/ Apple a miss.
Yesterday, Chinese President Xi Jinping’s endorsement of cheap, open-source AI at the World Artificial Intelligence Conference triggered a massive sell-off in US chip stocks. Apple rose over Nvidia – but the reality is both are near their obituaries being written in under five years. Both have zero manufacturing and an impossible to defend moat. Apple is particularly susceptible to being hit by device virtualisation.
This stock market reaction stems from fears that affordable Chinese ai alternatives will undermine the pricing power and market dominance of high-cost Western models.
Investors began aggressively exiting chip positions. The Nasdaq and S&P 500 experienced significant declines as the AI-driven tech rally faced a reality check. The bubble hasn’t bursted yet and it is hard to predict when it will. But the nature of the ai beast is repeatedly getting validated. Hyper over reality is expected in breakthrough technologies but this one is bigger than all earlier ones. Aided by ai of course.
Xi positioned China as a global provider of “AI for all,” championing open-source models as an alternative to proprietary, high-cost US models.
Dignitaries from 29 nations signed a China-led body to build AI capacity across developing countries. Chinese labs like DeepSeek, Alibaba, and Moonshot AI continue driving down inference costs.
Below a certain margin, U.S. businesses are not sustainable. They need someone like Indian resellers to improve their viability. So US made product Oracle and service TCS implemented.
Ai has not only disrupted TCS.
But it has also disrupted Oracle.
Or Intel and Microsoft.
Wintel is now history.
And that is the part very few understand. Nvidia is just a value added reseller for TSMC. And Microsoft is not going to become his value added reseller through Microsoft surface ultra type laptops.
Take data centres. The big monster data centres with hundreds of megawatts of power and billion litre water consumption don’t make sense.
What makes sense is gas cooled, water free, low power data centres.
