In a bid to replicate the success of international AI players, Indian startups often find themselves trying to build 'India's answer to OpenAI'. However, this approach is riddled with a fundamental flaw - the cost and complexity of frontier model training. A single run can cost billions of dollars, a financial burden that few Indian startups can afford.
But what if this thinking is misguided? India has its own unique problem spaces that require domain-specific data and solutions. Vernacular commerce, smallholder agriculture, and public health logistics are areas where the data is specific to the Indian context and can't be easily replicated from Silicon Valley. In these spaces, the focus should be on developing high-quality, relevant data that can drive meaningful impact.
India has a proven track record of creating technology public goods that are tailored to local needs. Aadhaar and UPI are prime examples of how designing solutions with Indian constraints in mind can lead to widespread adoption and success. Similarly, AI can be harnessed to create public goods that address specific Indian challenges.
The founders who understand this 'Indus Valley playbook' will be able to build durable companies that create real value for India. On the other hand, those who attempt to cosplay Silicon Valley will only end up building expensive demos that fail to make a lasting impact.
It's time for Indian AI startups to rethink their approach and focus on creating solutions that are uniquely suited to the Indian context. By doing so, they can create a sustainable and impactful presence in the global AI landscape.
