India’s AI boom may be quietly creating its next major dollar outflow problem.
As startups and enterprises rush to embed generative AI and large language models into everything from customer support to internal workflows, a growing share of their technology spending is flowing overseas through AI inference bills.
The New AI Bill: AI inference is becoming the biggest operating cost of AI-native products. Unlike classic software, every chatbot reply, workflow trigger or personalisation step can create a fresh bill, often tied to tokens or cloud compute. That means AI usage rises not just with customers, but with each interaction.
The Dollar Dent: For many early stage startups, AI inference already devours a big chunk of their cloud infrastructure budgets, a figure expected to rise even more going forward. With foundational models and GPU capacities tethered to dollar-linked pricing, Indian companies find themselves earning in Rupees but paying for their critical AI operations in dollars, which are at the mercy of market fluctuations.
The Value Leak: The concern is not just cost inflation, but also where the money flows. Much of India’s AI consumption ends up enriching overseas model providers and infrastructure owners rather than domestic compute ecosystems. As a result, critics are flagging that AI compute costs are starting to resemble a digital version of the import bill.
Engineering A Local Pivot: To counteract this economic drain, domestic players like Yotta, CtrlS and Reliance Jio are expanding GPU clusters and AI cloud offerings. Industry executives believe India has structural advantages in this space like a thriving developer ecosystem, domestic demand and lower operating costs for LLM makers.
Simultaneously, enterprises are already testing the local infrastructure for lower latency, better data residency and tighter control over costs. With the country projected to guzzle around 7 GW of AI compute capacity by 2030, will India remain dependent on foreign AI compute or will the domestic infrastructure step up? Let’s find out…
Quantum computing could eventually break today’s encryption. However, most current systems are not ready for that shift as existing quantum-safe options, especially fibre-based ones, remain limited by range and scalability. Enter Pramatra Space, a startup trying to fix this problem.
Reimagining Security: Founded in 2023, Pramatra Space is developing satellite-based Quantum Key Distribution (QKD) to create encryption keys that can be transmitted securely across long distances. By moving the problem into orbit, the startup aims to overcome the physical limits of terrestrial fibre networks and make global communication secure.
Chips In Orbit: The startup’s approach uses integrated photonic chips mounted on low Earth orbit (LEO) satellites to generate and transmit quantum keys to ground stations. Those keys are designed to be extremely difficult to intercept, giving enterprises and critical infrastructure a new layer of protection against future cyber threats.
Beyond Key Exchange: Pramatra is also planning to offer key management systems to secure digital assets and mission-critical networks. Alongside that, it is developing specialised chips for both satellite payloads and ground-based security modules, which could help it control more of the security stack and improve deployment flexibility.
As both governments and enterprises prepare for a post-quantum future, the startup is eyeing a piece of the global quantum cryptography market, which is projected to cross $4.6 Bn by 2030. So, can Pramatra build a secure backbone for India’s digital communication?
India’s watch market is no longer just Titan and Timex. From D2C microbrands to premium curators, a whole new generation of Indian watch brands is quietly building a market that didn’t exist a decade ago…
Source: Inc42 - Startups
