The Other Path, Why India Must Build AI That Solves Problems, Not Just Scales Like Silicon Valley
As global leaders gather at the AI summit in Delhi, a familiar script is playing out. The narrative centers on the US versus China, trillion-dollar valuations, compute races, and speculation about machines that may one day rival human intelligence. India appears in this story, but usually in a supporting role—a large market, a vast engineering workforce, a services hub. Rarely is it cast as a country that will shape the architecture of artificial intelligence itself.
This framing, as entrepreneur and scholar Vivek Wadhwa argues, reflects a hierarchy and a level of arrogance that places Silicon Valley and Beijing at the centre of technological destiny and treats everyone else as peripheral. It is a view that India must reject.
The Silicon Valley Script
Over the past three years, Silicon Valley has delivered a steady stream of grand pronouncements about AI’s future. AI will replace most human jobs within a decade. It will achieve general intelligence. It poses existential risks if not tightly controlled. And now, according to former Google CEO Eric Schmidt, the constraint may not be algorithms or talent but electricity itself.
Schmidt has argued that the US may need roughly 92 gigawatts of additional power to sustain AI’s trajectory—the equivalent of about 60 nuclear power plants. This projection is framed as proof of scale and inevitability. But it also carries a subtle message: only those who can marshal nuclear-scale power grids and limitless capital truly belong in the AI race. For a country like India, still balancing development priorities and energy transitions, the suggestion is that the decisive moves are being made elsewhere.
The message was reinforced when Nvidia CEO Jensen Huang declined to attend the summit in India, citing “scheduling conflicts.” He would not miss a similar gathering in China, where compute and semiconductor strategy are treated as instruments of national power. The contrast revealed how parts of Silicon Valley still view India: consequential, but not central.
The Flaw in the Scale Assumption
The central flaw in the dominant AI narrative is the assumption that scale equals progress. It assumes that tomorrow’s systems will simply require more GPUs, more data centres, and more electricity. This is a linear extrapolation, and history shows that linear extrapolation collapses at every major technological inflection point.
In the early 20th century, forecasters warned that expanding telephone networks would require millions of operators, because they assumed the future would simply be a larger version of the present. Technology rapidly erased that constraint. In 1943, IBM’s president predicted a global market for perhaps five computers. Today, billions of devices shape how we live, work, and think. As exponential technologies advance, they become smaller, faster, and cheaper—and their architecture changes.
AI will be no different. Already, specialised AI chips are reducing energy consumption per operation. Model compression and distillation techniques are shrinking large systems into smaller, efficient versions that retain core capabilities. Edge computing is moving intelligence closer to devices rather than concentrating it in hyperscale facilities. Open-source models are being optimised to run on standard hardware. The architecture is evolving rapidly.
The competition, Wadhwa argues, will not be won by those who build the largest models, but by those who design intelligence that works under real-world constraints and solves consequential problems.
India’s Comparative Advantage
This is where India’s strengths lie. India does not have the deepest venture capital markets or the most advanced semiconductor fabs. What it does have is unmatched experience in building digital systems at scale under constraint.
India built Aadhaar to give digital identity to more than a billion people. It built UPI to process billions of transactions at negligible cost. It launched ONDC to prevent monopolistic capture of digital commerce. These were architectural solutions built for India’s context, not copies of Silicon Valley platforms.
AI should follow the same logic. India does not need trillion-parameter models trained on the largest GPU clusters in the world. It needs models optimised for Indian languages, agricultural realities, healthcare gaps, and climate vulnerabilities. It needs systems that can run efficiently on accessible hardware and operate in low-bandwidth environments. It needs AI that strengthens farmers, doctors, teachers, and public servants.
A mid-sized model optimised for rural healthcare diagnostics may deliver more social value than a giant chatbot optimised for advertising engagement. An agricultural AI tuned to soil patterns and monsoon variability may double incomes without requiring nuclear-scale power expansion. Smaller, specialised systems designed for real constraints can outperform bloated general systems in human impact.
Different Incentives, Different Paths
Silicon Valley’s fixation with scale reflects its incentive structure. Venture capital rewards dominance. Public markets reward grand narratives. Size becomes a proxy for success. Bigger is treated as better because that is what the system pays for.
India operates under different incentives. It must solve for inclusion, affordability, and resilience. Its advantage lies in building systems that work for 1.4 billion people, not in chasing someone else’s benchmark.
This is not to romanticise constraint or to suggest that India should abandon ambition. It is to argue that technological progress does not follow a single linear path. There are multiple ways to advance, and the choice of path should reflect a country’s priorities, challenges, and strengths.
Leapfrogging the Centralised Model
Technology does not simply advance; it converges. Chips improve, sensors multiply, algorithms become more efficient, and architectures shift. India does not need to replicate today’s centralised infrastructure at national scale. It can leapfrog toward distributed, efficient intelligence designed for its own realities.
This is exactly what India did with telecommunications. While developed countries invested in expensive landline infrastructure, India leapfrogged to mobile. The result was faster, cheaper, and more inclusive connectivity. The same logic can apply to AI.
India also does not need to match the spending of the US or China. It needs to invest strategically, build indigenous capability, embed governance from the beginning, and focus on deploying the most useful systems rather than the largest ones.
Governance from the Start
One of the lessons from India’s digital journey is that governance cannot be an afterthought. Aadhaar succeeded not just because of technology but because of a legal framework that addressed privacy and security. UPI succeeded not just because of the platform but because of a governance structure that balanced innovation with stability.
For AI, governance must be embedded from the beginning. This means thinking about transparency, accountability, and fairness as design principles, not as compliance burdens. It means building systems that can be audited, explained, and contested. It means ensuring that AI serves people, not the other way around.
Conclusion: Uplifting Humanity, Not Inflating Valuations
The AI future will not be a single monolithic thing. It will be shaped by the choices made in different countries, by different actors, for different purposes. The competition will not be won by those who build the largest models, but by those who design intelligence that works under real-world constraints and solves consequential problems.
India has the talent, the experience, and the scale to be a leader in this kind of AI. It does not need to replicate Silicon Valley’s path. It needs to forge its own.
If it stays grounded in solving real problems and expanding opportunity, India will shape AI in a way that uplifts humanity rather than merely inflating valuations. That is a future worth building.
Q&A: Unpacking India’s AI Opportunity
Q1: What is the dominant narrative about AI, and why does it marginalize India?
The dominant narrative, shaped by Silicon Valley and Beijing, frames AI as a race to build the largest models, requiring massive compute clusters, nuclear-scale power grids, and limitless capital. India appears in this narrative as a large market and engineering workforce, but rarely as a shaper of AI’s architecture. This framing is reinforced when figures like Nvidia’s CEO attend summits in China but skip those in India, suggesting that only countries with the biggest infrastructure truly belong in the AI race.
Q2: Why is the assumption that “scale equals progress” flawed?
History shows that linear extrapolation collapses at major technological inflection points. Forecasters once predicted telephone networks would need millions of operators; technology erased that constraint. IBM’s president once saw a global market for five computers. As technologies advance, they become smaller, faster, cheaper, and their architecture changes. AI is already seeing this with specialised chips, model compression, edge computing, and open-source optimisation. The future belongs to those who design intelligence for real-world constraints, not just those who build the largest models.
Q3: What is India’s comparative advantage in AI?
India’s advantage lies in its unmatched experience building digital systems at scale under constraint. It built Aadhaar for a billion people, UPI for billions of transactions at negligible cost, and ONDC to prevent monopolistic capture. These were architectural solutions for India’s context. For AI, this translates into building models optimised for Indian languages, agricultural realities, healthcare gaps, and climate vulnerabilities—systems that work on accessible hardware in low-bandwidth environments. India’s incentive structure (solving for inclusion, affordability, resilience) differs from Silicon Valley’s (scale for dominance and valuation).
Q4: What does “leapfrogging the centralised model” mean for AI?
Just as India leapfrogged landline telephony to mobile, it can leapfrog today’s centralised AI infrastructure toward distributed, efficient intelligence. Instead of replicating massive GPU clusters, India can focus on edge computing, model optimisation, and systems designed for its realities. This approach requires strategic investment, indigenous capability building, and governance embedded from the start—not matching US or China’s spending, but deploying the most useful systems at scale.
Q5: Why is governance particularly important for India’s AI path?
India’s digital successes (Aadhaar, UPI) succeeded not just because of technology but because of legal frameworks that addressed privacy, security, and governance. For AI, governance must be embedded from the beginning—transparency, accountability, and fairness as design principles, not compliance burdens. This ensures AI serves people, can be audited and contested, and builds trust. As India forges its own AI path, getting governance right will be as important as getting the technology right.
