73% think India will
lead AI. 72% are
already behind.

The $500 billion report nobody's
reading carefully enough

On Wednesday, IBM and IndiaAI released From Promise to Power — a survey of 1,500 Indian executives across industry sizes and leadership levels. The headline the press ran with: AI could contribute more than $500 billion to India's economy by 2030. That's a real number. But it's the third paragraph that deserves the attention.

One in three employees currently has the AI literacy that businesses say they need. By 2030, that figure needs to be one in two — which means training 350 million workers in under four years. For context, that's more people than the entire workforce of the United States, European Union, and UK combined.

The report's most honest line, buried in the middle: "India's ability to scale AI is shaped not by the sophistication of the models but by the readiness of enterprise data and infrastructure." That's the whole thing. Fancy models don't matter if your data is a mess and your cloud costs are 3× what a startup in San Francisco pays.

Fair enough. But participation and leadership are different things. The $500 billion is what happens if we close the execution gap. It's not a projection of current trajectory. The report, to its credit, is honest about that. The coverage, less so. Full report →

Android stops being an OS.
Google I/O is Tuesday.

The Google I/O keynote is on 19 May — five days from now. The pre-event Android Show dropped on Thursday and it's bigger than a routine update. Google's VP of Android, Sameer Samat, said the quiet part out loud: "We're transitioning from an operating system to an intelligence system."

What that means in practice: Gemini Intelligence, a new AI layer across Android 17, will handle multi-step tasks directly inside your apps — booking parking, pulling information from your emails, filling forms, completing purchases — without you manually navigating between them. The phone becomes less a tool you operate and more an agent that acts on your behalf.

Also announced: a new laptop category called Googlebook, Android XR glasses, deeper Gemini integration into Chrome and Android Auto, and Quick Share expanding to Apple devices via QR codes. The full I/O keynote is likely to add more AI-specific announcements — models, Veo, developer tools.

Tune in Tuesday. Google has kept leaks unusually tight this year, which usually means they have something worth protecting.

AI that speaks before
you ask. All three labs.
Same week.

ChatGPT Pulse launched in September 2025 — a personalised morning briefing delivered to Pro users before they've typed a word. Last week, Anthropic followed with Orbit, connecting to Gmail, Slack, GitHub, Calendar, Drive, and Figma to send proactive summaries. This week, Google pre-announced Gemini Proactive Assistance ahead of I/O.

Three labs. Eight months. Same direction. AI that initiates rather than responds.

There's a less charitable reading of why this is happening simultaneously. The $20/month subscription model has a ceiling: users who don't open the app for two consecutive weeks tend to cancel. Proactive AI manufactures daily engagement — it's a retention mechanism disguised as a feature. That doesn't make it bad. It just means you understand what you're signing up for.

The tools themselves are genuinely useful. The question is whether you want your AI thinking about you when you haven't asked it to.

1.5 lakh developers.
A very short window.
Starting now.

On 12 May — three days ago — NASSCOM launched AI Code Sarathi, a national upskilling programme targeting 150,000 developers for production-grade AI coding capability. Not AI awareness. Not a certification badge. The goal is moving developers from knowing what AI is to actually building with it in a production environment.

The timing is not accidental. LinkedIn data shows AI-specific job demand in India grew 59.5% year-over-year. That's the demand side. AI Code Sarathi is an attempt to move the supply side fast enough to meet it.

The broader picture from the IBM report: India needs 350 million AI-literate workers by 2030. Code Sarathi is targeting 150,000 developers. The scale gap is enormous. But 150,000 developers who can actually ship AI products is not nothing — those are the people who build things the remaining 350 million use.

14% of downloads.
2% of the revenue.
India in two numbers.

Wispr Flow is an AI voice dictation app from San Francisco. Between October 2025 and April 2026, it was downloaded 2.5 million times globally. India accounted for 14% of those downloads — the second-largest market in the world, behind only the United States. India also contributed 2% of in-app purchase revenue during the same period.

That 7:1 usage-to-revenue ratio is not unique to Wispr Flow. It's not even unique to AI. It's the defining structural challenge of every consumer technology company operating in India. Indian users adopt early, use heavily, and pay reluctantly at global price points. The product works — 70% of Indian users are still on it after 12 months. The monetisation doesn't, at $12/month.

Wispr Flow's response: India-specific pricing at ₹320/month, down from ₹1,000+. Long-term target: ₹10–20/month. CEO Tanay Kothari told TechCrunch he wants "every single person in the country to be able to use Wispr Flow." That's not a product decision. That's a choice to accept that India may never be a revenue market at global prices — but might matter for everything else.

The reason voice AI is hard in India specifically: not just price sensitivity, but linguistic complexity. India has 22 official languages, hundreds of dialects, and a communication pattern — Hinglish — that switches languages mid-sentence in ways that break most models. Wispr Flow hired two full-time linguistics PhDs to fix this. That's the kind of commitment it takes to actually serve this market, not just claim it.