In a recent interview highlighted by The Economic Times in its article “Why Sarvam’s Vivek Raghavan Calls Data in Motion the Most Valuable Asset for AI Startups,” Sarvam co-founder Vivek Raghavan argued that the competitive edge in artificial intelligence is shifting decisively away from static datasets toward real-time data streams.
Raghavan’s perspective reflects a broader evolution within the AI ecosystem. While early advances in machine learning were powered by large, curated datasets, he contends that such resources are increasingly commoditized. What now distinguishes high-performing AI systems, he suggests, is access to continuously generated, context-rich data that reflects live user interactions, operational environments, and rapidly changing conditions.
This concept of “data in motion” encompasses dynamic inputs such as conversational exchanges, user behavior patterns, transaction flows, and sensor outputs. Unlike static datasets, which eventually become outdated or saturated, real-time data allows AI systems to adapt, learn, and improve continuously. For startups, this capability can translate into faster iteration cycles, sharper personalization, and more resilient models in production environments.
Raghavan also points to the strategic implications for younger companies navigating an increasingly competitive AI landscape dominated by large technology firms. While incumbents may possess vast repositories of historical data, startups can carve out defensible positions by capturing and leveraging unique streams of live data. This creates what he describes as a compounding advantage: the more an AI system interacts with users, the more refined and valuable its underlying data becomes.
However, the emphasis on data in motion introduces new technical and ethical challenges. Managing real-time data pipelines requires robust infrastructure, low-latency processing capabilities, and sophisticated feedback loops. At the same time, companies must navigate concerns around data privacy, consent, and security, particularly as real-time systems often involve sensitive or personally identifiable information.
The argument also carries implications for how investors evaluate AI ventures. Rather than focusing solely on model architecture or training scale, attention may increasingly shift toward a startup’s ability to acquire, process, and retain high-quality data streams. In this context, distribution and user engagement become as critical as technical innovation.
Sarvam itself, which is focused on building AI tailored to Indian languages and contexts, is positioned to benefit from this approach. By embedding its systems within real-world applications and interactions across diverse linguistic and cultural settings, the company aims to generate precisely the kind of dynamic data that Raghavan describes as most valuable.
As the AI sector matures, the notion that static datasets alone can sustain long-term advantage appears to be weakening. Raghavan’s emphasis on data in motion underscores a shift toward systems that are not just trained once, but continuously evolving. For startups seeking to differentiate themselves, the race may no longer be about who has the most data, but who can harness it as it unfolds.
