India's $68B AI Bet: Building Skyscrapers on Quicksand

February 19, 2026 (2m ago)

February 19, 2026 — Today, Prime Minister Narendra Modi stood before a constellation of tech CEOs—Sundar Pichai, Sam Altman, Dario Amodei—and unveiled India's "MANAV" vision for AI: Moral, Accountable, National sovereignty, Accessible, and Validity-based.

It sounded inspiring. The kind of framework that gets quoted in think tank reports and LinkedIn posts.

There's just one problem: the summit itself collapsed under its own weight.

When Vision Meets Reality

Let's start with what actually happened at the India AI Impact Summit 2026:

• 70,000 attendees showed up. The Wi-Fi didn't.
• Bill Gates pulled out last minute amid renewed Epstein scrutiny, dealing a "fresh blow to a flagship event already marred by organisational lapses," per Reuters.
• Prototypes were stolen from exhibitor booths. Yes, stolen. At India's flagship AI summit.
• Chatham House published an analysis titled "Do AI summits work?" and concluded: "little progress on international governance is expected."

This isn't a minor logistics hiccup. It's a metaphor.

You can't run an AI summit without reliable Wi-Fi. You can't build an AI superpower without reliable infrastructure.

The MANAV Vision: Noble Goals, Missing Foundations

Modi's MANAV framework calls for:

• Moral and ethical AI systems
• Accountable governance
• National data sovereignty
• Accessibility and democratization
• Validity and legal standards

These are admirable principles. In fact, they're the right principles.

But here's what the summit didn't address:

1. India has fewer than 300 skilled AI researchers

The UK has more. France has more. India spends 0.7% of GDP on R&D, compared to China's 2.68%.

When Sridhar Vembu (Zoho founder) watched Trump advisor Sriram Krishnan tell India to "leverage" American AI infrastructure instead of building its own, he responded bluntly:

"This is why brain drain is costly and we must fight hard to retain the next generation of talent in India."

India graduates hundreds of thousands of engineers annually. The best ones leave. Not for lack of ambition—for lack of competitive research infrastructure, funding, and career paths.

2. The power grid can't handle what's coming

India's power grid faces a "silent exit" risk as AI data centers are set to explode 4x by 2030.

The Indian Express reports:

"AI, especially during the prompt phase, is highly unpredictable, which would make forecasting and scheduling challenging for data centres."

Translation: even if you build the data centers, you can't reliably power them.

TechCrunch noted that India faces "structural challenges, including access to reliable power and water for energy-intensive data centers, underlining the execution risks."

You know what AI training runs need? Uninterrupted gigawatt-hours of electricity. You know what India's grid delivers? Variability.

3. There's a 10:1 talent gap

The Times of India reported in 2025:

"For every 10 AI roles, there is 1 engineer."

Think about that. India has the demand. It just doesn't have the supply.

Cloud computing roles face a 55-60% demand-supply gap. Cybersecurity? Equally severe shortages.

When your Finance Minister openly acknowledges an "AI talent gap" while industries are "struggling with a shortage of skilled talent pool," you're not positioning for leadership—you're scrambling to keep up.

4. Previous AI summits produced "few enforceable outcomes"

Reuters put it plainly when covering this summit:

"Previous global AI summits...produced voluntary corporate pledges and governance declarations, though critics said they produced few enforceable outcomes."

The Paris AI Summit pivoted to "boosterism instead of reckoning with the gravity" of AGI risks. The Seoul AI Summit focused on frontier model safety commitments—voluntary ones.

India's summit? More of the same. Big numbers ($68B in commitments through 2030), big names (Google, Microsoft, Amazon), big visions (MANAV).

But what's actually enforceable? What's measurable? What happens if those commitments don't materialize?

The $68 Billion Question

Here's the number everyone's citing: $68 billion in AI and cloud infrastructure investment from Google, Microsoft, and Amazon through 2030.

That's massive. It's real money.

But investment announcements aren't infrastructure. They're promises.

India has "long project cycles, land acquisition, regulatory approvals, and grid integration planning" ahead, per Startup News. Execution will determine credibility.

And here's where it gets uncomfortable: India is trying to "compress years of AI infrastructure build-out into a much shorter time frame" while dealing with:

• Aging grids
• Slow permitting
• Regulatory complexity
• Talent outflow
• Funding shortfalls

You can announce $68B. You can't will it into existence.

What India Gets Right (And What It Doesn't)

What India gets right:

• Positioning as a scaling ground. With advanced economies dealing with regulatory congestion, India could become the place where next-gen AI infrastructure gets built.
• The demographic advantage. Young, tech-savvy population. Massive engineering graduate pipeline.
• The ambition itself. You don't become an AI leader by being modest.

What India doesn't get right:

• Confusing summits with strategy. A four-day event with celebrity CEOs isn't a substitute for decade-long infrastructure investment.
• Ignoring the fundamentals. You can't leapfrog power grid reliability. You can't outsource talent retention. You can't MANAV your way past a 10:1 skills gap.
• Treating brain drain as inevitable. When your top AI talent builds the future for San Francisco instead of Bangalore, sovereignty is just a buzzword.

The Uncomfortable Truth

Here it is, plainly:

AI summits don't build AI leadership. Execution does.

You want ethical AI? Start by building an ecosystem where Indian researchers don't have to leave for competitive salaries and compute access.

You want accountable governance? Start by running a summit where Wi-Fi works and exhibits don't get stolen.

You want national sovereignty? Start by retaining the talent that builds foundation models instead of watching them join OpenAI.

India has the potential. It always has. What it lacks isn't vision—it's the unglamorous, grinding work of:

• Stabilizing power grids
• Funding university AI labs at scale
• Creating competitive research careers
• Streamlining data center approvals
• Building domestic GPU access
• Retaining talent through infrastructure, not rhetoric

The Bottom Line

PM Modi's MANAV vision is the right framework. Ethical, accountable, sovereign AI—these are values worth building toward.

But you can't build a skyscraper on quicksand.

And right now, India's AI ambitions are standing on:

• A power grid that can't handle 4x data center growth
• A talent pipeline that's leaking its best people to the US
• An R&D budget that's 1/4 of China's
• Infrastructure promises that face "long project cycles" and "execution risks"

The question isn't whether India wants to be an AI leader. It's whether India is willing to do the hard, boring, unsexy work to become one.

Summits make headlines. Infrastructure makes futures.

We'll see which one India actually builds.

What do you think? Is India's AI ambition achievable, or are we watching another cycle of big promises and small execution? Tell me on Twitter/X.