AI's future lies in affordable, accessible intelligence, not just smarter models

If you spend enough time around Artificial Intelligence (AI), it's easy to believe the only thing that matters is who has the smartest model. Every few months there's a new leaderboard. Better reasoning, bigger context windows, higher benchmark scores. It's an incredible engineering achievement, but I don't think it will define the next couple of years. The biggest shift in AI won't be another model that's slightly smarter. It will be the cost of intelligence falling so dramatically that AI becomes something we stop thinking about.

History tells us what happens when powerful technology becomes cheap. We don't use less of it. We use it everywhere. Computers got cheaper, so every home got one. Storage got cheaper, so we stopped deleting photos. Bandwidth got cheaper, so streaming became the default. AI is following the same path. Stanford's 2025 AI Index found the cost of GPT-3.5 level performance fell from $20 per million tokens in late 2022 to $0.07 by late 2024, more than 280x in less than two years. That's one of the fastest cost curves in technology.

Most people assume that trend means AI companies will spend less. I think the opposite happens. When intelligence becomes cheap enough, we stop rationing it. Instead of asking AI five questions a day, it will help hundreds of times without us noticing. Your inbox gets summarised before breakfast, meeting prep is ready before your first call, and followup emails are drafted before you've opened your laptop. AI stops being a chatbot you visit occasionally and becomes something that's simply there whenever you need it. Most users won't know, or care, which model answered. They'll expect AI to be fast, reliable and affordable.


That shift matters because, despite all the excitement, we're still living inside a bubble. If you spend your day in Silicon Valley, it feels like everyone uses AI. They don't. The world has more than 8.2 billion people. ChatGPT alone crossed 1 billion monthly users in June 2026, and that's still a fraction of humanity. Around 2.2 billion people don't have reliable internet access, and countless others have never had a meaningful AI experience. We're much earlier than people think. The next billion AI users will come from AI solving real problems in everyday life, not from benchmark debates on X.

That's why I think India will be one of the most important AI markets in the world, starting now rather than a decade out. India has repeatedly shown how transformative technology can reach massive scale once it becomes affordable. UPI made digital payments frictionless for hundreds of millions of people. Affordable mobile data brought an entire generation online. Aadhaar turned digital identity into infrastructure for over a billion people. ChatGPT already has more than 100 million weekly users in India alone. AI has the chance to follow the same playbook. Imagine every student having a tutor in their own language, every doctor having an assistant, every entrepreneur having access to capabilities that only large enterprises could afford a few years ago. That's a much bigger opportunity than another benchmark victory.

It's also why I'm so passionate about democratising AI. At Fireflies, we've always believed that every time intelligence gets cheaper, more people should benefit. As AI gets better, it should also get more accessible. I'm hoping we get both smarter and cheaper AI. But for the next few years, accessibility is the bet I'd make. History won't remember who built the smartest model this quarter. It will remember who made intelligence available to everyone.

(The author is co-founder and CEO, Fireflies.ai. He will be speaking at The Economic Times World Leaders Forum in New Delhi)