Sarvam AI — India's full-stack sovereign AI lab — closed the first tranche of a $300M Series B on June 15, led by HCLTech ($150M), with Bessemer, Khosla Ventures, and Peak XV joining; cash goes toward compute and training for its next frontier model targeting coding, agentic, and cybersecurity workloads. For builders: this concentrates senior ML and infra talent in Bengaluru into one well-funded lab, driving up comp bands fast — lock your AI hiring offers now, and watch Sarvam's open-source Sarvam-30B/105B MoE models (Apache 2.0 on Hugging Face) as production-ready multilingual base layers.
Pramaana Labs (founded 2025) raised a $27M seed led by Khosla Ventures, with Accel, Nexus, Premji Invest, and Unbound participating, to build AI systems that output checkable, auditable proof alongside answers — targeting tax, healthcare, cybersecurity, and financial compliance use cases. For builders in regulated verticals: 'verifiable AI' is emerging as a distinct product category in India, signalling that enterprise buyers are prioritising explainability and audit trails over raw LLM capability.
In March 2026, RBI released its Payments Vision 2028 roadmap, emphasising cross-border UPI expansion, single-window licensing for authorised persons, and NPCI now having the authority to revise P2M transaction limits dynamically without seeking fresh RBI approval — P2P remains capped at ₹1 lakh. For fintech builders: the shift to NPCI-led dynamic limit-setting means product roadmaps for high-value merchant flows (B2B, healthcare, education, travel) can now be planned with greater regulatory flexibility, while the cross-border UPI push opens real GTM paths across 8+ live countries.
Indian AI startups pulled in $3.94B in Q1 2026 alone — more than 6x the full-year 2025 total — anchored by Neysa's $1.2B GPU infrastructure round and Sarvam's sovereign model funding; the IndiaAI Mission has onboarded 38,000 GPUs offered to startups at ~42% below market rates, with a target of 100,000 public GPUs by December 2026. For builders: subsidised sovereign compute means Indian AI startups can now train and fine-tune models without burning VC cash on cloud bills — a structural cost advantage that Western rivals lack, making India the most capital-efficient geography to build foundation-layer AI products.
Get this brief in your inbox every morning — free
One edition a day, every number sourced and dated. Pick your industries after signing in.
Subscribe free