When AI CEOs Won't Hold Hands: Inside the India Summit

There's a photo op from the AI Impact Summit in New Delhi that tells you everything about where we are with AI right now. More than a dozen tech leaders stand with Prime Minister Modi, arms raised in that universal gesture of unity and celebration. Except two of them—Sam Altman and Dario Amodei—conspicuously refuse to hold hands.

The internet, naturally, did its thing. Someone analyzed the tape frame by frame (shoutout to Bef Jesus) and determined Amodei was the one who pulled away first. But honestly? That detail doesn't matter as much as what the moment represents: the massive gap between AI as a "global public good" (the summit's stated theme) and AI as a ruthlessly competitive race between companies whose CEOs literally won't touch each other.

The Summit That Tried to Reframe Everything

This was the first time the AI Impact Summit happened in a developing country. Previous editions were in the UK, France, South Korea—places already deep in the AI race. Holding it in India was supposed to signal something different: that AI development can't just be a conversation among rich countries.

UN Secretary General António Guterres went hard on this point, writing: "The future of AI cannot be decided by a handful of countries or left to the whims of a few billionaires. AI must belong to everyone. AI must be accessible to everyone. AI must benefit everyone. AI must be safe for everyone."

He followed up with a call for a global fund to build "skills, data, affordable computing power, and inclusive ecosystems everywhere." It's one of the first times we've heard world leaders explicitly frame affordable AI access for the Global South as a priority, not just an afterthought.

India itself showed up ready to play. Reliance Industries and another major player each committed over $100 billion for local data centers. The Indian government added $1.1 billion to the pot. These are not symbolic numbers.

But then you had that photo op. And Bill Gates canceling his keynote because of renewed Epstein file scrutiny. And the main event: watching Altman and Amodei deliver wildly different speeches while a viral chart suggested Anthropic is on track to overtake OpenAI in revenue by mid-year.

The Performance Art of AI Leadership

Amodei's speech was... not great. He read from his iPhone (which people really noticed), delivering what multiple observers described as generic talking points he's said a thousand times before. One engineer on X wrote: "The aura loss is crazy. I take back everything good I said about anthropic."

Altman, meanwhile, was eloquent and engaged, discussing how AI's fundamental uncertainty intersects with democracy, social contracts, and job displacement. His pitch: keep deploying iteratively, let people access each layer of technology as it develops. Classic Altman—optimistic but not naive, acknowledging problems while pushing forward.

In a CNBC interview, he threw some shade at the AI-job-loss narrative: "I don't know what the exact percentage is, but there's some AI washing where people are blaming AI for layoffs that they would otherwise do. And then there's some real displacement by AI of different kinds of jobs."

Here's the thing though: Sean Wang (Swix) articulated what a lot of people were feeling. He wrote that the summit was India's moment on the global stage, potentially an inflection point for 1.5 billion people figuring out their place in the AI economy. "And yet the powers that be decisively demonstrated that nothing will change. They care more about bad photo ops and hobnobbing with celebrities than they care about the builders that are supposed to drive the Indian AI economy forward."

I keep thinking about that tension. You have genuine, massive investment commitments. Real policy discussions about AI as infrastructure, not luxury. And then you have... whatever that hand-holding moment was.

Meanwhile, In Corporate America

While world leaders were doing their thing in New Delhi, corporations were quietly making AI adoption much more concrete and much less optional.

Walmart's earnings call focused heavily on their AI shopping assistant, Sparky. CEO John Furner reported that about half of Walmart's online customers have used it, and those who do order 35% more than those who don't. That's not a rounding error—that's a business model shift. Walmart's calling it a move from "traditional search to intent-driven commerce," which is corporate-speak for "we're using AI to figure out what you want before you fully know it yourself."

Amazon's approach is more surveillance-flavored. They're using an internal system called Clarity to track AI tool adoption across teams. Not just among engineers—supply chain, operations, everywhere. Employees are being asked to provide examples of how they've "accomplished more with less" and "force multiplied using AI" while "reducing or not growing headcount." The framing is incredibly transparent about what's actually happening here.

But Accenture? Accenture went full mandate. They're explicitly tying AI adoption metrics to promotions for senior managers. Use the tools or don't advance. Last year their CEO said people who fail to adopt AI workflows would be "exited from the company." This week's email made clear: training's over, now it's table stakes.

The Financial Times noted this is becoming a pattern across big consulting. Senior partners are resisting more than junior staff—they're "more set in their ways," requiring a "carrot and stick approach." One employee said they'd "quit immediately" if the policy affected them. Another described Accenture's tools as "broken slop generators."

The Adoption Paradox

There's a comment from someone on X (Hedgy Markets) that captures a common reaction: "If these tools were actually useful, people will just use them. You don't need to track logins and tie them to promotions. The fact that companies are resorting to this tells me adoption isn't happening organically."

It's not a cynical take, but I think it misses something crucial. The pattern that keeps showing up in surveys isn't that the tools don't work—it's that people don't have time to learn them. You need time to figure out how to make AI save you time, but most companies don't create that space. They just pile AI tools on top of existing workloads and expect organic adoption.

The second most common complaint? The tools approved for work are way behind what people use at home. "At home I'm using Claude Opus 4.6, at work I have a terrible old version of Copilot." That creates justified resentment—being forced to use inferior tools while better ones exist.

Accenture's mandate makes a certain sense given their business model. If you're selling AI transformation to clients, your own people need to actually know the tech. But their stock is down 45% over the past year, which suggests the "AI transformation" pitch isn't landing as hard as they'd hoped.

Two Conversations, Same Planet

So here's where we are: world leaders in India talking about AI as a global public good while the CEOs on stage won't hold hands. Massive infrastructure investments alongside bitter commercial rivalry. Companies mandating AI use while employees complain the tools don't work.

The Altman-Amodei moment was uncomfortable, sure. But it was also clarifying. We're having two completely different conversations about AI at the same time. One is about democratization, access, and treating AI as shared infrastructure for human flourishing. The other is about market share, revenue curves, and competitive moats.

Both conversations are real. Both will shape what happens next. But they're increasingly hard to hold simultaneously, and that hand-holding refusal was just the moment the tension became visible.

India committed billions. The UN called for global funds. Companies are forcing adoption. And two CEOs stood on stage, arms raised but not touching, each representing a different vision of what AI should become.

The question isn't which vision wins—it's whether we can build systems that serve both, or if that's fundamentally impossible.

—Yuki Okonkwo