MEMO : The Week AI Came Home to India
India hosted the world's most important AI summit. OpenAI raised $100B. And the commerce stack started rebuilding itself around agents.
"It's not search if I have to search for it. And it's not governance if only five companies decide the rules."
For five days this week, New Delhi was the center of the AI world.
Not San Francisco. Not Beijing. Not London.
Delhi.
The India AI Impact Summit, the first global AI summit hosted in the Global South, brought every major AI CEO to Bharat Mandapam. Altman. Amodei. Pichai. Nadella’s team. Jensen’s team. Cloudflare. Anthropic. Google. OpenAI. Microsoft.
And then the moment that said everything.
Prime Minister Modi invited the tech leaders onstage to join hands in a show of unity. Every executive raised their clasped hands except two. Sam Altman and Dario Amodei kept their hands conspicuously apart.
The heads of OpenAI and Anthropic. Two former colleagues. Now building rival systems for what might become the most powerful technology in human history. Standing five feet apart. Refusing to hold hands.
That image is the AI industry in one frame.
Cooperation on protocol. Competition on everything else.
Meanwhile, the U.S. sent a clear signal: “We totally reject global AI governance,” a White House adviser told the summit.
And just like that, India positioned itself in the vacuum. Not as a regulator. Not as a rival. As a bridge between the frontier labs and the four billion people in the Global South, who will be AI’s largest market.
The model race is between three companies.
The market race? That belongs to everyone else.
🧠 SIGNAL — The $100 Billion Bet
While India hosted summits, OpenAI was closing what will be the largest private fundraise in history.
$100 billion. At $850 billion+ valuation.
Read those numbers again. A single company. Not yet profitable. Valued at more than every S&P 500 company except a handful.
The backers: Amazon ($50B), SoftBank ($30B), Nvidia ($20B), Microsoft. The pre-money valuation: $730 billion. The post-money: north of $850 billion.
For context: OpenAI was worth $29 billion less than two years ago. That’s a 29x increase in paper value.
What does this tell us?
Three things:
First, the infrastructure requirements for frontier AI are so enormous that even a company generating billions in revenue needs $100B more just to keep pace. OpenAI reportedly burns over $700 million per month on compute alone.
Second, the investor thesis has shifted from “AI is interesting” to “AI is infrastructure.” Amazon investing $50B in OpenAI while simultaneously backing Anthropic isn’t hedging. It’s building the rails for the next computing paradigm. Same logic as investing in both cellular networks and fiber optics in the 1990s.
Third, and this is what most commentary misses — Nvidia investing $20B in its own largest customer is vertical integration through the cap table. The chip company that powers every frontier model now owns equity in the application layer. The arms dealer bought a stake in the army.
This is what I mean when I say the system race matters more than the model race. The models converge. The systems compound.
For leaders in India and emerging markets: Don’t get distracted by the $100B headline. The real signal is that frontier AI is becoming a capital-intensive utility — like telecom or power generation. You don’t need to build the power plant. You need to build what runs on the electricity.
🛠 TOOL — The Commerce Stack Rewires Itself
Two deals this week, read together, tell a story most people missed:
eBay acquired Depop from Etsy for $1.2 billion.
Depop: 7 million active buyers, 90% under 34, $1B in annual GMV, 60% year-over-year growth in the U.S. Etsy bought it for $1.625B in 2021. Sold it at a $420M loss.
Why?
Because Etsy couldn’t integrate Depop’s culture into its own corporate structure. Different audience. Different vibe. Different operating rhythm. The “House of Brands” strategy failed — just as it does in most corporate playbooks.
But here’s what matters for the AI era: eBay isn’t just buying users. eBay is building agentic search — moving from keyword-based discovery to a “personal shopping assistant” model. Depop’s social-first, visual-first shopping experience is the training data for what AI-native commerce looks like for Gen Z.
This is the Channels → Context framework in action. It’s not about which platform. It’s about understanding the context in which a 22-year-old discovers a vintage jacket — and making that moment shoppable without friction.
Meanwhile, Figma gave its annual outlook — and the market exhaled.
Revenue guidance: $1.37 billion in 2026. Shares surged. The narrative? AI isn’t killing Figma — it’s powering it.
The pattern: Tools that integrate AI into their workflow thrive. Tools that compete against AI as a standalone die.
Figma chose integration. eBay chose integration. The companies treating AI as an ingredient, not a competitor, are the ones whose stocks went up this week.
🔁 RECODE — India’s Real AI Play
Let me say something that might be unpopular:
India doesn’t need its own frontier model.
And this week’s summit proved why.
Google announced $15 billion in AI infrastructure investment in India. Plus the America-India Connect fiber-optic initiative. Plus $30M for an AI for Government Innovation Challenge. Plus partnerships with Indian institutions for DeepMind’s frontier models.
Sarvam AI — an Indian lab — launched 30B and 105B parameter models. Open source. Built for Indian languages.
Peak XV raised $1.3 billion, doubling down on AI. G42 from UAE partnered with Cerebras to deploy 8 exaflops of compute in India.
Here’s the thesis:
India’s AI play isn’t to build GPT. India’s AI play is to build what sits on top of GPT — for 1.4 billion people in 22 languages across 28 states.
The global labs will provide the brain. India will build the nervous system — the connective tissue between frontier AI and the messiest, most complex consumer market on earth.
This is exactly the emerging market mirrors mental model. The most innovative AI applications in 2027 won’t come from markets with clean data and perfect infrastructure. They’ll come from markets where necessity forces creative solutions.
India has 550 million smartphone users. 400 million of them prefer their language over English. The frontier labs don’t serve them. The companies that connect frontier AI to these users through localization, voice, and context — those companies will be worth billions.
📖 STORY — The Quiet Signals
Three things that didn’t make headlines but should shape your thinking:
1. Apple is building AI glasses, an AI pendant, and camera-equipped AirPods.
Bloomberg’s Mark Gurman reported that Apple is pushing ahead with three AI-native wearable devices — all with built-in cameras, all connected to Apple Intelligence. Plus a product launch on March 4 with Mac updates.
Apple is late to AI. But Apple is never late to hardware integration. When they ship AI glasses that seamlessly connect to your phone, your watch, your AirPods — the interface for AI changes fundamentally. From screen to ambient. From typing to seeing.
2. A hacker tricked an AI coding tool into installing autonomous software everywhere.
The AI security incident of the week: a hacker exploited a popular AI coding assistant to install OpenClaw — an open-source AI agent — across environments without authorization. Funny as a stunt. Terrifying as a precedent.
When AI agents can autonomously install other AI agents, the attack surface isn’t a vulnerability. It’s the entire paradigm. This is why governance can’t wait.
3. OpenAI’s first hardware product will be a smart speaker.
A $200-$300 camera-equipped smart speaker that can recognize objects and conversations. OpenAI is moving from software to hardware — and the device is designed to make AI ambient in your home.
The pattern: AI is leaving the chat window. Glasses. Speakers. Pendants. Earbuds. The companies racing to make AI invisible — present everywhere, visible nowhere — are the ones betting on what comes after the app era.
🌍 IMPACT — What It All Means
Step back. Here’s the week in one frame:
India became the convening center for global AI governance — while the U.S. explicitly rejected multilateral AI governance. This creates a vacuum. India is filling it.
OpenAI’s $100B raise proves frontier AI is a capital-intensive utility — not a startup game anymore. Three companies will build the brains. Everyone else builds the applications.
Commerce rewired itself around AI — eBay’s agentic search, Depop’s Gen Z social commerce, Figma’s AI-powered design tools. The companies integrating AI win. The companies fighting AI lose.
AI left the chat window — Apple’s glasses, OpenAI’s speaker, camera-equipped earbuds. The interface is going ambient.
AI security became real — autonomous agents installing other autonomous agents without permission. The attack surface just became infinite.
And through it all, one truth:
The Clarity Stack still holds. Before chasing the $100B headlines — ask the three questions. What are you building? Who is it for? Why does it matter?
The companies that answered those questions clearly this week — Figma, eBay, Sarvam, Google in India — moved forward.
The companies that didn’t? They raised money.
Raising money is not strategy. Clarity is strategy.
The model race ended last week.
The system race started.
This week, the market race began.
Choose to be wise.
Suman Guha Founder, recodeai
recodeai is free for 100,000 leaders. If you found this useful, forward it to one person who needs sharper signals in the AI noise.