E2E Networks Limited — Q3 FY25 earnings call

Call held 22 Jan 2025

Management summary

E2E Networks reported strong year-on-year growth in Q3 FY25 across revenue, EBITDA, and PAT, driven by advancements in cloud infrastructure and AI/ML focus. However, the company experienced a quarter-on-quarter decline in revenue and profit, attributed to bursty training workloads from large customers and muted demand in December. Management emphasized an agile CAPEX strategy and deepening strategic partnerships like with L&T to capture larger enterprise opportunities and expand its accelerated cloud infrastructure.

Highlights

  • Total revenue stood at INR 41.6 crores, witnessing a substantial growth of 73.7% YoY, but a decline of 12.6% QoQ.

  • EBITDA for the quarter was INR 24.6 crores, growing 119% YoY, but declining from INR 31.4 crores QoQ.

  • EBITDA margin for the current quarter was 59%, demonstrating a growth of 1223 bps YoY.

  • PAT was reported at INR 11.6 crores, showing a growth of 108% YoY, but a decline of 4.3% QoQ.

  • Diluted EPS was 7.03, an 86.5% increase YoY.

  • Cumulative CAPEX deployed reached INR 201.7 crores till December 2024.

  • INR 150.8 crores of raised funds utilized till Q3 FY25, with a balance of INR 1334.9 crores.

Key financials

  1. Revenue ₹41.6 Cr +73.7%YoY
  2. EBITDA ₹24.6 Cr +119%YoY
  3. EBITDA Margin 59%
  4. PAT ₹11.6 Cr +108%YoY
  5. PAT Margin 27.8%
  6. Diluted EPS ₹7.03 +86.5%YoY

What they filed

Q1 FY27: revenue up 336.1%, net profit up 1566.7% against the same quarter last year.

₹ Cr · quarterly
Line itemQ2 FY25Q3 FY25Q4 FY25Q1 FY26Q2 FY26Q3 FY26Q4 FY26Q1 FY27
Revenue48 42 33 36 44 −8%70 +67%96 +191%157 +336%
EBITDA31 25 13 11 18 −42%40 +60%58 +346%118 +973%
Net profit12 12 14 -3 -13 −208%-6 −150%6 −57%44 +1567%
How to read this

₹ crore, as filed. The percentage beside a figure is the change against the same quarter a year earlier — never the quarter before, which would make every seasonal business look like it collapses and booms each year.

Guidance & targets

Capacity

  • Chennai Data Center Operationalization Capacity · Q4 FY25 · Medium confidence Operational in next quarter
    We are trying to do it as soon as possible. I guess like during the next quarter, it would get operationalized whether we will be able to deploy all the services in the first quarter itself, like remains to be seen.

    — Tarun Dua, Managing Director

Capex

  • CAPEX Deployment Capex · next several quarters · Low confidence Just-in-time deployment as much as possible
    So, we are anticipating that like majorly based on the pipeline we will evaluate and based on that like we will kind of like do just-in-time deployment as much as possible over next several quarters.

    — Tarun Dua, Managing Director

  • Blackwell GPUs Acquisition Capex · as and when it becomes available · Medium confidence Build a significant amount of capacity
    And yes, absolutely, we do intend to build a significant amount of capacity on the Blackwell range as and when it becomes available.

    — Tarun Dua, Managing Director

  • Hopper Range CAPEX Capex · short term · Low confidence Reactive CAPEX
    There could be reactive CAPEX in the Hopper range. It would depend on the outlook that is coming from the sales conversations that we are having.

    — Tarun Dua, Managing Director

Revenue Mix

  • Training vs Inference Revenue Ratio Revenue Mix · medium and long term · Medium confidence 50:50 or 60:40 in favor of training
    ultimately it's going to be somewhere the ratio eventually over the medium and long term, I think it should become more like 50:50 or maybe 60:40 in favor of like training.

    — Tarun Dua, Managing Director

Risks & concerns

  • Quarter-on-Quarter Revenue Decline

    medium

    Revenue declined 12.6% QoQ due to churn and downscaling of bursty training deployments from large customers, concentrated in Q3, and muted demand in December.

    Management acknowledged

  • Underutilization of H100/H200 GPUs

    medium

    The H100 and H200 GPUs are currently underutilized, impacting immediate revenue generation from new capacity.

    Management acknowledged

  • Longer Sales Cycles

    medium

    Sales cycles have grown, contributing to the difficulty in quickly recovering unutilized infrastructure to higher utilization rates.

    Management acknowledged

  • ARPU Reduction

    medium

    ARPU has reduced due to some larger customers on the training side having churned or downscaled their deployments.

    Management acknowledged

  • Impact of US GPU Export Rules

    low

    Management believes the impact of new US GPU rules would be assessed over a two-year timeframe rather than immediately, as current volumes are not hitting limits and exceptions exist.

    Analyst downplayed

Areas of evasion (1)

  • Share price commentary

Q&A highlights

2 direct
Reason for Quarter-on-Quarter Revenue Drop Direct
We currently have a fairly small scale and compared to that, some of our larger customers, their deployments were quite large. Now, training workloads by their very nature are bursty... all of these got concentrated during the Q3. And we have also seen somewhat muted demand for the end of this quarter, December typically is a slow month.

Reveals the impact of bursty workloads from large customers and seasonal demand slowdown on the company's relatively small scale.

Asked by Aastha

H100/H200 GPU Utilization and CAPEX Deployment Timeline Partial
The H100 and H200, are they completely utilized or they are underutilized as of now? No, they are underutilized as of today... Very very hard to commit to kind of like spend that money in a market where the infrastructure... they get like upgraded version. So, it is very, very important not to kind of like build a lot of inventory... So, that has always been our strategy that like always be able to invest for every version.

Highlights current underutilization of advanced GPUs and management's cautious, reactive CAPEX strategy to avoid obsolescence and maintain flexibility, despite having significant cash reserves.

Asked by Keshav

Training vs Inference Revenue Percentage and Future Outlook Direct
The major concentration of the revenue obviously has been training, will continue to be training... ultimately it's going to be somewhere the ratio eventually over the medium and long term, I think it should become more like 50:50 or maybe 60:40 in favor of like training.

Clarifies the current dominance of training workloads in revenue and provides a long-term directional target for the split between training and inference, indicating future business model evolution.

Asked by Ashwin Kedia

2 min read 6 chapters

Detailed narrative

Q3 FY25 Financial Performance Overview

E2E Networks reported a mixed Q3 FY25, with strong year-on-year growth but a notable quarter-on-quarter decline. Revenue grew 73.7% YoY to INR 41.6 crores, but fell 12.6% QoQ from INR 47.6 crores. EBITDA increased 119% YoY to INR 24.6 crores, achieving a 59% margin, but was down from INR 31.4 crores in the previous quarter. PAT also saw a 108% YoY increase to INR 11.6 crores, with a 27.8% margin, despite a 4.3% QoQ decline. Diluted EPS stood at 7.03, up 86.5% YoY.

Strategic Partnerships and Infrastructure Expansion

The company deepened its strategic partnership with L&T, aiming to leverage L&T's data center expertise and E2E's AI compute infrastructure to offer scalable solutions to enterprises and government. This collaboration is expected to open new revenue streams and drive growth in high-demand AI services. E2E is also expanding its data center capacity from 4.2 megawatts to nearly 10.2 megawatts, supported by recent fundraising, and is establishing a second facility near Chennai, expected to be operational in Q4 FY25.

GPU Deployments and Utilization

E2E Networks has significantly expanded its GPU infrastructure, with cumulative deployments reaching nearly 700 H100 GPUs, 256 H200 GPUs, and around 700 non-H100/H200 GPUs. However, management acknowledged that the H100 and H200 GPUs are currently underutilized. The QoQ revenue decline was partly attributed to the bursty nature of large customer training workloads and muted demand in December, which led to underutilization of this capacity.

AI/ML Workload Dynamics and Revenue Mix

The company's AI/ML platform, TIR, is designed to streamline AI/ML workloads, offering a superior price-to-performance ratio. Management stated that the major concentration of revenue is currently from training workloads, which are inherently bursty. While difficult to classify precisely, the long-term outlook for the training-to-inference revenue ratio is projected to normalize to 50:50 or 60:40 in favor of training, as the AI market matures.

CAPEX Strategy and Funding

E2E Networks raised INR 1484.9 crores through equity shares in 2024, utilizing INR 150.8 crores till Q3 FY25, leaving a balance of INR 1334.9 crores. The CAPEX deployment strategy is described as 'agile' and 'just-in-time,' reacting to demand rather than being predictive, to avoid building inventory of rapidly evolving high-tech equipment like GPUs. The company intends to build significant capacity on the upcoming Blackwell range of GPUs as they become available, with reactive CAPEX also possible for the Hopper range.

IndiaAI Mission and Market Outlook

The company views the IndiaAI Mission as a net positive for the entire AI industry in India, expecting it to expand the overall market regardless of specific beneficiaries. E2E has qualified in the technical evaluation for a government AI project tender and is awaiting the financial bid results. Management maintains a very positive outlook for compute infrastructure and AI services in India, believing the market will continue to grow over the medium to long term.

This is an AI-generated summary of a publicly available earnings call transcript.
View original transcript

More Information Technology earnings calls