Skip to content

Blog

Insights on marketing, growth, and AI.

Analysis, frameworks, and field notes for founders, marketing leaders, and teams putting AI to work.

Landlord or tenancy - Should we celebrate the AI data center?

India didn't win an AI data center. It signed up to be the landlord. Meta just leased its first India AI campus: 168MW in Jamnagar. Reliance builds it, owns it, runs it. Meta pays the rent. Read the headline and it's a sovereignty win. Read the structure and it's a tenancy deal. Reliance carries the capex, the land, the power, the operational risk. Meta keeps the parts that appreciate: the model, the IP, the customer. And it can walk away from the lease. We're supplying cheap renewable power and a coastline. They're keeping the layer that compounds. I'm not knocking it. At 168MW, it's real industrial muscle. But owning the building isn't the same as owning the economy that runs inside it. The question for Indian boards isn't "are we building data centers?" It's "are we building anything that appreciates?" Landlord or architect: which one is India actually becoming?

Continue reading: Landlord or tenancy - Should we celebrate the AI data center?

THE PRE-IPO SCORECARD: RATING ZEPTO'S AI READINESS BEFORE THE DRHP

I built a pre-DRHP scorecard — eight categories, 40 points, India-specific. Zepto today: 24/40. The customer-facing AI work is strong; the operating-discipline work is unproven externally. Fraud and rider-onboarding KYC are the lowest-scoring categories. AI governance disclosure is next. Both can be closed before the DRHP lands — most quick-commerce filings globally have not, and that gap becomes a discount. The three questions the roadshow has to answer crisply: forecast error vs. baseline, share of AI decisions with no human override, board owner of AI governance. A team that needs the slide pack has a 20/40 story.

Continue reading: THE PRE-IPO SCORECARD: RATING ZEPTO'S AI READINESS BEFORE THE DRHP