AI-generated · cited to primary sources · not investment advice · How we research
E2E Networks
Our verdict on E2E Networks isn’t the consensus take — see where we landed, and the one risk the bull case glosses over.
See the verdict — free →Does management do what it says?
The company has already reached a capacity of nearly 3,900 GPUs by the end of Q1 FY26, surpassing the earlier target of ~3,700. (2 exceeded, 3 met across 5 tracked commitments)
“2025 - Capacity Expansion GPU’s Capacity reaching to ~3900 GPU”
The company has achieved its target of having over 100 engineers on staff. (2 met across 2 tracked commitments)
“Yes, anywhere between 75% to 90% utilization is what we are expecting.”
The company has utilized a total of INR 8,842.13 Mn in Q1FY26 alone from the preferential issue funds, significantly accelerating its GPU procurement and infrastructure spend beyond the initial FY25 assignment. (3 exceeded, 1 missed, 1 met across 5 tracked commitments)
“So, broadly in line, where we feel that basically 70% to 80%, 70% to 75% kind of EBITDA should be possible. So, going out on a limb I would not go to 70%, but let's say, 65% to 75% should be doable from EBITDA margin perspective with the capacity that we are expecting to kind of convert to revenue by March exit month in the next year, in this year.”
The partnership is yielding actual enterprise customer conversions and a specific 1-year GPU service contract starting Jan 2026. (1 in progress across 1 tracked commitment)
“Target Enterprise Customers with Sector-Specific Cloud Solutions”
Path for expansion through AI Lab As A Service (AILaaS) for educational institutions. — target: Scalable, cost-effective solution for educational institutions
“AI Lab As A Service: AILaaS is a scalable, cost-effective, and customizable solution designed to help educational institutions”
See the full cited Management analysis of E2E Networks
How durable is the business?
The technology moat is expanding with the launch and integration of the TIR platform, which now supports advanced models like Llama 4 and Mistral, and provides a full-stack environment for AI training and inference. (2 expanding)
“TIR – End-to-End AI infrastructure solution... Unifies containerized compute GPU acceleration, and integrated ML tooling”
The company significantly accelerated its infrastructure build-out, increasing GPU capacity from ~1,000 to 3,700 units, with a target of 10,000+ GPUs in the next 2-3 years. (5 expanding)
“Current operational GPU capacity: 3,900+ units... Expected capacity by March 2026: ~ 5000+ total GPUs”
Average Revenue Per User (ARPU) for top 500 customers showed high volatility in the latest quarter, dropping from Q2 peaks, though still significantly higher than FY24 levels. (1 shifted, 1 contracting, 1 expanding)
“Quarterly ~ARPU For Top 500 customers (INR) Q4FY25: 5,80,000”
EBITDA margins for Q4 dropped to 40% from a steady-state 60% as the company prioritized free trials (POCs) for large enterprise customers to secure future long-term contracts. (1 shifted)
“And many of the current large unicorns have used E2E Networks compute infrastructure when they scaled up themselves from start-up to unicorn stage.”
While India remains the primary market, the acquisition of Jarvis Lab assets marks a shift toward global expansion, targeting international customers with a self-service, high-margin model. (1 shifted)
“The Jarvis Lab acquisition of assets was completed in this quarter... this is a technology-centric move that rapidly positions us to scale up in the global market.”
See the full cited Business Model analysis of E2E Networks
Where does growth come from?
Capacity expansion is accelerating significantly; the company moved from ~1,000 Hopper GPUs to 3,700 total GPUs recently and is now deploying an additional 2,048 H200 GPUs, with a long-term target of 10,000+ GPUs. (5 accelerating across 5 signals, 2 leading indicators)
“But some of them are already running their inference workload with us and many of the new ones will hopefully are already working with our customers through our partners for the inference workloads, and we continue to see the acceleration of this trend.”
The monthly revenue runrate shows a clear upward trajectory, with a significant breakout in December 2025 reaching INR 280 Mn, indicating strong progress toward the higher MRR targets. (1 accelerating across 1 signal, 1 leading indicator)
“TIR – End-to-End AI infrastructure solution. Pre-configured environments that enable faster experimentation and model deployment”
The company has aggressively scaled its GPU capacity, reaching approximately 3,700 Cloud GPUs by FY25, supported by a massive jump in capital expenditure to INR 8,700 Mn. (2 accelerating, 2 new trend, 1 steady across 5 signals, 2 leading indicators)
“And we continue to hold our target of being able to hit somewhere close to INR35 crores to INR40 crores of MRR around March 2026 or thereabouts.”
The L&T partnership is entering a new phase of active leveraging to pursue larger enterprise deals and international business, moving beyond the initial setup phase. (1 new trend, 2 steady across 3 signals)
“And we have already seen some progress in terms of actual conversions happening amongst the enterprise customers through one of our largest partnerships with L&T”
A major server outage in Mumbai during December highlights a potential risk to customer trust and reliability, though management claims the financial impact is small. — Service Reliability (Mumbai Outage): Non-material revenue impact expected
“Also, Tarun, in December, we faced a major outage in the Mumbai servers... we don't expect to see a material revenue impact over the medium term and the long term.”
See the full cited Future Growth analysis of E2E Networks
What could break the thesis?
The risk is INTENSIFYING as EBITDA margins dropped significantly from 60% to 40% this quarter due to high costs associated with idle capacity and non-revenue generating trials (POCs). (5 intensifying, 2 high-severity)
“Finance Cost 47 ... 135.0% [QoQ]”
INTENSIFYING: While the Mumbai outage was previous, the company is now facing significant technical delays in its new Chennai facility, which was expected to be live but is currently in 'Work in Progress' (CWIP) due to cluster-related technical issues. (1 intensifying, 1 easing)
“Also, Tarun, in December, we faced a major outage in the Mumbai servers. So just wanted to seek some clarification on your end and if there's any major client loss due to the outage?”
The risk is INTENSIFYING as operational revenue actually declined 12.6% YoY to INR 361 Mn despite the massive increase in hardware capacity (Capex). This suggests a temporary mismatch between the timing of capacity deployment and revenue realization. (1 intensifying, 3 easing, 1 stable)
“Expected capacity by March 2026: ~ 5000+ total GPUs”
While the L&T partnership helps de-risk deals, the company's revenue is becoming increasingly tied to large strategic contracts, which could lead to lumpy earnings if renewals or new contracts are delayed. [CONCENTRATION]
“IndiaAI Mission contracts: ₹265 Cr combined ... New L&T contract: ₹8.49 Cr”
INTENSIFYING. While annual revenue is up, the Monthly Revenue Runrate and ARPU for top customers show a sharp decline in the final two quarters of FY25. Monthly revenue peaked at INR 165 Mn in Sep-24 but fell to INR 112 Mn by Mar-25. Quarterly ARPU also dropped from a peak of INR 8,60,927 in Q2FY25 to INR 5,80,000 in Q4FY25. (1 intensifying)
“Monthly Revenue Runrate: Sep-24: 165, Dec-24: 113, Mar-25: 112; Quarterly ~ARPU For Top 500 customers: Q2FY25: 8,60,927, Q4FY25: 5,80,000”
See the full cited Risk analysis of E2E Networks
E2E Networks analysis by filing period
Read the full thesis on E2E Networks
AI-generated informational research only. ThesisLoop is not investment advice, a stock recommendation, or a guarantee of returns.