The LLM market is saturated. Wise firms should build platforms instead.
They call it Fomo. Fear of missing out is a powerful force. China has it, so built DeepSeek in response to the brigade of large language models (LLMs) emanating from Silicon Valley. DeepSeek won the battle, if not the war. Built at a fraction of the cost of ChatGPT, DeepSeek is smaller, more efficient – and vastly cheaper to run – than anything from the Californian fleet.
The Chinese interloper triggered a gloomy spell of introspection that stretched from Santa Clara to the Statue of Liberty. If Asia could build a worthy rival to the Valley’s flagship with pocket change, and in the blink of an eye, what did it say about US tech’s ability to win the global race? The Nasdaq had an immediate answer: the shimmer of its prize stocks dulled to a suddenly unimpressive hue.
Sometimes Fomo bears fruit. But you can have too much of a good thing. The success of China in entering the LLM arena has caused rampant envy across the Himalayas. Its ancient adversary, India, now wants a piece of the prize. There is the rattling of sabers; the prospect of a New Delhi new-build designed to shake San Francisco and shudder Shenzhen.
India’s government officials advocate billion-dollar investments in data centers and computing infrastructure. Its academics demand sovereign LLMs. Industry groups nationwide lobby for more public-private collaborations in model training. But they are all focusing on yesterday’s problem, fighting yesterday’s war. The world is awash with LLMs. The market has become saturated. Whatever niches there were have been overfilled. Enough already with the Fomo. Today’s opportunity lies elsewhere.
India should avoid building its own LLMs. Rather, it should embrace the foreigners – and run with them. There is no sense in recreating what already exists. Wise firms – in India, and globally – are best advised to build on what’s already there. AI’s new frontier is not more models. It is platforms.
In my book, From Incremental to Exponential, I outline how platforms laid the foundations for Silicon Valley’s global dominance. Platforms became the fathers of trillion-dollar companies. By allowing others to build atop their infrastructure, they fostered entire digital ecosystems that attracted developers, locked in users, and fueled unstoppable network effects.
Innovators from shores thousands of miles from west-coast America or south-coast China can replicate this success in the AI era. They can use existing AIs as mere foundations for new platforms that can attract mass adoption. They can construct digital cities on the roads, railroads and subways shaped from LLMs. A good platform provides tools, reach, and scale. The best platforms become the backbone of entire industries. And the next wave of platforms will be powered by AI.
The global story of surveying the foundations then raising the bar is the essence of technological revolution. The two superpowers of our time – Apple and Microsoft – never dirtied their hands with the groundwork. Rather, they identified the infrastructure and fashioned epoch-shifting solutions from the world around them. That they are now among the most valuable companies on the planet is no coincidence: if you build it, they will come.
The fight for the next LLM has become a scrap with no winner. India,
for one, would be wise to keep a safe distance. Let others distract themselves with Fomo. There are glittering prizes to be won – by using the spoils of others’ LLM feuds as a launchpad for the AI platforms of tomorrow.
