88% of AI Money Goes to the US — and Sarvam Shows India Has an Alternative

88% of global AI investment stays in the United States. Meanwhile, India produces its first pure AI unicorn with Sarvam, and infrastructure startups raise $1.8 billion in 48 hours. Three stories that aren't about bigger models — they're about who controls the means of AI production.

🇺🇸 88% of AI Money Goes to the US — and the Rest of the World Wonders If This Is Sustainable

Crunchbase has published an analysis that puts numbers to what many suspected: the AI investment boom is not global. It's American. So far in 2026, US startups have captured nearly 80% of all global seed-through growth-stage financing. But when you filter for AI alone, that figure jumps to 88%: roughly $319 billion, most of it going to just two companies — OpenAI and Anthropic.

To put it in perspective, before the generative AI boom, American companies never exceeded 50% of global investment. The shift is structural. China, with $33 billion so far this year, already exceeds its 2025 total. The UK has raised $16.5 billion. But no other country comes close to the capital concentration that Silicon Valley generates. The uncomfortable question is whether this is a legitimate competitive advantage or a financial bubble with an American accent.

🇮🇳 Sarvam AI Becomes a Unicorn: India Bets on AI Sovereignty

HCLTech has led a $234 million funding round in Sarvam AI, India's largest pure-play AI startup. The investment, which values the company at $1.5 billion, also includes Nvidia, Prosperity7, Bessemer Venture Partners, and existing investors Khosla Ventures and Peak XV Partners. Nvidia's involvement is particularly telling — the chip giant doesn't invest in every startup that asks.

Sarvam's plan is ambitious and concrete: use the capital to train its next frontier model focused on agentic capabilities, code generation, and cybersecurity. But the most interesting part isn't the model itself — it's the vision of Indian digital sovereignty. HCLTech, as a strategic partner, will get preferential access to develop industry-specific language models and expand Sarvam's multilingual capabilities, a critical factor in a country with 22 official languages and hundreds of dialects.

🏗️ $1.8 Billion in 48 Hours: AI Infrastructure Eats the Pie

Baseten — an inference infrastructure company — is closing a $1.5 billion round that values it at up to $13 billion. On the same day, General Intuition, a New York startup training world models on billions of video game clips, is negotiating a $300 million round. Neither builds foundation models in the GPT or Claude sense. Both sell infrastructure. Together they've raised $1.8 billion in two days.

This reflects a thesis that has been maturing all year: base models are commoditizing faster than anyone expected. With open-source models from Meta, Mistral, and DeepSeek reaching quality thresholds where enterprises no longer need to pay proprietary API premiums, value is migrating to the layers around the model — the training data and world representations that determine what agents can do, and the inference infrastructure that determines how cheaply and reliably they can do it.

🌐 The Thread That Connects Them

Three stories from seemingly separate compartments — finance, geopolitics, technology — but they tell the same story: AI is reshaping who holds the capital, who builds the infrastructure, and who controls the data. The US concentrates the money, but that concentration raises questions about sustainability. India shows there's an alternative path built on national sovereignty and enterprise-government collaboration. And the infrastructure startups remind us that when the dust settles, what matters won't be who has the biggest model — but who can run it cheapest and most reliably.

The practical lesson is clear: AI money isn't going where there's the most intelligence — it's going where there's the most capacity to distribute that intelligence at scale. If you're building on AI, don't obsess over the model — obsess over the delivery layer. That's what investors are paying for.