A US export control on Anthropic's newest model accidentally made the strongest case for India's own AI infrastructure in months. The timing wasn't planned. But it was noticed. One big story, five fast hits, one free AI-agent workflow you can build this weekend.
On 9 June, Anthropic released Claude Fable 5. Three days later, the US Department of Commerce ordered it pulled for every foreign national, anywhere in the world, effective immediately. Anthropic had no way to verify nationality in real time. So it shut Fable 5 — and its sibling Mythos 5 — down for everyone, everywhere, that same day.
The trigger: Amazon researchers found a jailbreak. A specific prompt got Fable 5 to identify software vulnerabilities and, in one case, write code showing how one could be exploited. The government treated that as serious enough for emergency export controls.
Here's what didn't make the US tech coverage: Sarvam's $234M Series B closed on 15 June — three days into the blackout. TechCrunch, Business Standard and Bloomberg all opened their coverage of that round the same way: it landed the same week the US government proved it can switch off a frontier model for an entire country, overnight, with zero warning to the people using it.
The Anthropic story ended well enough. Over 18 days, the company worked with the government, trained a new safety classifier that blocks the specific technique in over 99% of tries, and got the controls lifted on 30 June. Fable 5 came back on 1 July. Mythos 5 stays on a tighter leash — still limited to a small number of vetted US organisations under Project Glasswing.
Sarvam isn't a Fable 5 substitute — it isn't trying to be. The Anthropic story is over; the bigger question it raised isn't. Every developer or company building on a foreign frontier model is carrying a dependency risk they probably haven't priced in. The Hindu ran an op-ed on exactly this on 1 July — sovereign AI as a UPSC-syllabus policy question, not a startup pitch. That's a signal about how seriously this is now being taken in Delhi. We covered last week why HCLTech's $150M into Sarvam looks smart as a business bet. After this fortnight, it also looks smart as an insurance policy.
Five things worth knowing from the past week — headline, the gist, and why it matters for India.
India has the largest AI-skilled workforce outside the US — but only 5–10% of those 2 million have "advanced" skills, per Nasscom/SHRM data. The rest can use AI tools. Very few can build, evaluate or maintain an AI system in production.
Why it matters: That gap is why AI hiring stays brutal even as overall tech layoffs continue. "I use ChatGPT" and "I can build an AI agent" are different sentences in a 2026 interview — and the gap between them is where the real salary premium sits.
Indian tech services firms already earn $10–12B in AI revenue. Nasscom's projection for the agentic wave is $300–400B in additional addressable spend by 2030. But fresh data from multiple research firms shows roughly 25% of enterprise AI agent projects are in production — the rest are pilots dressed as products.
Why it matters: When a vendor comes to you with an agentic pitch this year, ask for a production reference. The gap between the $300B opportunity and 1-in-4 production rate is where most of the vendor promises live.
The Act's obligations for general-purpose and high-risk AI systems activate on 2 August 2026. A new independent scientific panel begins advising the AI Office. The key clause most Indian firms haven't checked: the Act applies to any organisation that offers AI systems or services to users in the EU, regardless of where the provider is based.
Why it matters: If you have EU clients, EU users on a SaaS platform, or AI in a product sold to European companies — this is live in four weeks. Not planning to comply is a position, but it should be a conscious one.
Coverage of Sarvam's round consistently flags that Krutrim — India's first AI unicorn — "recently pivoted to AI cloud services" as of May 2026 and is no longer focused on its own frontier models. Sarvam is now the clearest flag-bearer for the "build our own" camp in India.
Why it matters: Worth knowing before you frame a roadmap around "Indian alternatives." The list of Indian companies actually training and maintaining frontier models is shorter than the press coverage implies.
Baseten, a US AI inference provider, closed a $1.5B Series F in late June at roughly a $13B valuation — one of the largest AI infrastructure rounds on record. The company helps companies deploy and scale model inference in production.
Why it matters: Useful scale check. Sarvam's entire $234M round — which we're calling India's biggest sovereign AI bet — is a rounding error next to what US AI infrastructure is raising. The gap in capital isn't the whole story, but it's part of the honest picture.
Fast Hit #1 says the gap is advanced AI skill, not enthusiasm. Here's how to close it — this weekend, for ₹0.
The detail I keep coming back to from this fortnight: nobody outside the US had any say in the Fable 5 shutdown. Not the developers who'd shipped product on it. Not the companies whose workflows went dark. The order took effect the same day it arrived. That's not a hypothetical risk — it happened, it lasted 18 days, and it resolved in a way that had nothing to do with anyone outside Washington and San Francisco.
Sarvam's $234M didn't happen because of the blackout. But the framing of every single major coverage piece — from TechCrunch to Bloomberg to Business Standard — was shaped by it. That's worth sitting with the next time someone asks whether sovereign AI is a marketing line or a real strategic need.
If this was useful, or if you're going to build the Make.com workflow this weekend, forward this to someone who'd find it worth five minutes. That's the whole distribution plan.
See you next Saturday.
The same five-minute format every week: the one thing that matters, a few fast hits, and one workflow you can actually use — all with India context and real INR pricing.
Free forever · No spam · Unsubscribe any time