₹4,073 crore market cap. Stock price ₹2,024. Three-month return of –37.6% that looks like a ski slope in Gulmarg. Quarterly revenue just exploded to ₹70.02 crore, a 68.3% YoY jump, while quarterly PAT calmly reported a ₹5.70 crore loss, waving at profitability like “next season, bro.” Operating margins? A juicy 56.6% in Q3 FY26, because GPU clouds print operating profit before depreciation shows up like an uninvited auditor. ROCE sits at 8.08%, ROE at 5.71%, and promoter holding is 40.35% with 60.4% pledge, which is the financial equivalent of saying “I believe in the business but also in leverage.” Sales over the last five years compounded at 45.6%, profits earlier compounded hard, but TTM profits are down –122% thanks to aggressive capex and depreciation. This is not a sleepy IT services firm. This is a hyperscale AI cloud sprinting ahead while accounting standards jog behind with a whistle.
2. Introduction – From Small Cloud to GPU Mafia Boss
E2E Networks started as a cloud infrastructure provider when “cloud” still sounded like weather. Today, it markets itself as an AI-first hyperscale GPU cloud, which in India basically means: “AWS prices but desi compliance, sovereign comfort, and GPUs that are actually available.” While global hyperscalers sell capacity like airline tickets during Diwali, E2E decided to buy the planes.
The company now sits right at the intersection of three powerful forces: India’s AI ambition, government-backed IndiaAI compute demand, and enterprises suddenly realizing that training models on laptops is not a career-safe decision. Add a strategic stake from L&T, thousands of NVIDIA H100/H200 GPUs, and data centers popping up faster than coaching institutes in Kota, and you get a company that looks futuristic on PowerPoint and painful on the P&L.
But here’s the twist: E2E’s losses are not from weak demand. They are from too much ambition too fast. Depreciation, interest, and upfront investments are doing bhangra on the income statement, while revenue is still warming up its vocal cords. This article dissects whether this is visionary suffering or just very expensive optimism.
3. Business Model – WTF Do They Even Do? (Explained Without Buzzword Bingo)
At its core, E2E sells compute. Not generic CPU compute that runs Excel, but serious GPU compute that trains large language models, runs GenAI inference, and makes data scientists feel important.
The crown jewel is the TIR AI Platform, a full-stack AI/ML ecosystem. Think Jupyter notebooks for development, vLLM and LoRA for efficient model serving, vector databases for embeddings, integrations with WandB, and resume-training features so your model doesn’t cry if power trips. On top of this sits a self-service public cloud, allowing customers to spin up Linux, Windows, and GPU machines hosted entirely in India.
Then comes the Sovereign Cloud Platform, launched in 2025. This is aimed at government bodies and enterprises that want AWS-like features but with Indian data residency, compliance, and privacy. E2E claims deployment in 15 days, which in government time is basically teleportation.
Finally, AI Lab-as-a-Service targets educational institutions, giving them GPU access without selling kidneys. This creates early ecosystem lock-in: today’s students, tomorrow’s enterprise
