Profit as Infrastructure: How Eternal’s Margin Engine Funds Its AI Bet
The Keys Were Never Stolen – And That’s Exactly What Should Worry Us
The most dangerous failures in a trust system don’t announce themselves as breaches. They look like an ordinary Tuesday. Last week’s eMudhra episode proved the point: eight private keys in a listed digital trust company’s infrastructure were so weak that researchers could reconstruct them from publicly available certificates. No heist, no insider, no zero-day. The keys were simply born wrong – generated with small, guessable values during a 2025 key-generation ceremony, where a flaw quietly minted a weak set alongside the strong ones. Google, Mozilla, Apple and Microsoft have moved to block the certificates; the market answered with a lower circuit.
I’ve spent two decades building systems where the invisible plumbing decides the outcome, and this incident is a masterclass in an uncomfortable architectural truth: security is a genesis problem before it is a defence problem.
We pour budgets into perimeters, threat intelligence and anomaly detection – the runtime of security. But cryptographic material has a birth certificate. If entropy fails at generation time – a flawed RNG, a reused seed, a misconfigured ceremony – every downstream control inherits that weakness. You cannot patch randomness; you can only revoke and re-mint. That asymmetry is what should keep architects up at night: one bad ceremony, and remediation isn’t a hotfix, it’s a planetary revocation coordinated across every major browser and operating system.
Three imperatives follow:
1. Treat key ceremonies as production systems, not rituals. Observable, hardware-backed (HSMs, true RNGs), multi-party controlled, with verifiable randomness. If you cannot audit how a key was born, you have no business trusting it. Yes, ceremony-grade rigour is slow and expensive – but the failure asymmetry demands over-investment at genesis.
2. Engineer for cryptographic agility. Rotating trust anchors across a distributed fleet is now a first-class architectural requirement, not a nice-to-have. Treat this episode as the dress rehearsal for post-quantum migration: when that day arrives, the survivors will be the organisations that already know how to re-mint trust at scale.
3. Audit your trust supply chain. Enterprises consume certificates the way consumers consume packaged food – assuming the factory was clean. Make entropy provenance and key-generation hygiene a formal procurement question, not an assumption.
There’s a second signal from the same week worth reading alongside it: Eternal unveiling Dark, a privacy-first AI platform built on fine-tuned open-weight models. Ignore the launch fanfare and look at the architecture – the market is beginning to price confidentiality as a design-time property rather than a compliance checkbox. Open weights plus private inference is where enterprise AI is heading, and both stories whisper the same thing: trust is migrating from the perimeter into the core of the stack.
And there is a distinctly Indian lesson here. Our Digital Public Infrastructure – eSign, digital signatures, eKYC – runs on exactly these trust anchors. Millions of MSMEs, from Guwahati to Kochi, sign tenders and invoices on these rails, and when an anchor wobbles, the last mile feels it first. eMudhra says its root CA and Indian DSC/eSign infrastructure were unaffected, and that containment matters. But zoom out: in the same week, institutional investors subscribed their Jio Platforms IPO quota over 9x – capital is making a historic bet on India’s digital rails. Bets of that size only hold if the trust beneath them is beyond doubt.
Where I’d direct the next budget conversation:
- Inventory your cryptographic estate. You cannot rotate what you cannot see.
- Demand key-generation transparency from every vendor in your chain.
- Put rotation drills in the disaster-recovery plan, not the wiki.
- Design AI systems for data minimisation now – privacy-first is becoming table stakes, not differentiation.
Code can be patched. Trust has to be re-issued. In the architecture of a digital nation, that distinction is everything.
About the Author: Sanjeev Sarma is the Founder Director and Chief Software Architect at Webx Technologies. With a core focus on Generative AI integration, Cloud-Native Scalability, and Enterprise Software Architecture, he has spent over two decades driving digital transformation across Northeast India and beyond. Beyond his corporate leadership, Sanjeev is deeply invested in shaping the future of the IT industry. He serves as an Industry Expert on the Board of Studies for Assam Don Bosco University’s School of Technology, advises state technology committees, and actively mentors emerging tech startups at STPI. He brings a unique, dual perspective of high-level enterprise execution and future-ready academic curriculum development.
