E2E Networks Ltd Q2 FY2024 Concall Summary

Key Takeaways

• Growth story intact, capacity is the bottleneck: Revenue growth is a function of consistently increasing monthly recurring revenue, but the buildout of infrastructure to capture AI/ML demand takes time and has faced global supply-chain delays.
• PAT margin boosted by accounting change: The 26.85% Q2 PAT margin benefited from a ~INR 3 Cr impact due to a shift in depreciation method from WDV to SLM, a non-operational tailwind.
• Bold medium-term growth target set: Management anticipates 40% to 50% revenue growth over the next couple of years, driven by a focused AI/ML infrastructure play and a 50% to 60% cost advantage versus hyperscalers.
• CAPEX cycle is front-loaded and flexible: A ₹150 to ₹200 crore hardware deployment is planned over 12 to 15 months, with 30% to 35% of orders already placed, but management retains the ability to scale down if demand tapers before funds are committed.
• What to watch next quarter: The pace of delivery for the 30% to 35% of CAPEX orders already placed, with 70% to 75% expected this quarter, and any early signs that the 6-month sales cycle from deployment to uptake is accelerating.

Growth Engine & Capacity Buildout

• Revenue mechanics: Growth is driven by consistently increasing monthly recurring revenue, with a customer base that is increasingly defined by workload rather than company size, as larger customers adopt the self-service platform for production infrastructure.
• Capacity lag: Infrastructure buildout takes time and has been delayed by global supply-chain issues for primary products; post-creation, the entire sales cycle from deployment to significant uptake is about 6 months, with full uptake taking 5 to 12 months.
• PoC flywheel: With more capacity built, E2E can run a larger number of proofs of concept, though the time for a customer to move from onboarding to production remains constant.
• CAPEX plan: ₹150 to ₹200 crores of hardware deployment is planned over the next 12 to 15 months, with 30% to 35% of orders already placed; about 70% to 75% of those orders should be delivered within this quarter.
• Funding stance: For current plans, debt and internal cash flows make the CAPEX possible without raising additional equity; management does not see raising equity on a very immediate basis.
• ARR conversion: If the full ₹200 crores of CAPEX is deployed, it should result in additional annualized recurring revenue of ₹120 to ₹140 crores over and above the investing revenue.

Tarun Dua: "We want to ensure that we are able to cater to that demand while being flexible and nimble. So, which means that if we see that demand taper before we have committed to spend that money, we can obviously scale that down."

Competitive Positioning & Pricing

• Cost advantage: E2E is typically 50% to 60% cheaper overall compared to large hyperscalers, a gap enabled by a lower cost structure from not doing everything hyperscalers do, such as edge computing or private 5G.
• Focused strategy: The company is very focused on AI/ML workloads and infrastructure built for AI enablement and training, unlike hyperscalers with a broad service portfolio, allowing a smaller, more efficient player to have a lower cost despite hyperscalers' scale advantages.
• Pricing stability: E2E does not change pricing on the majority of SKUs unless there is a cost mismatch; pass-through licensing SKUs change only if the backend license changes, and once a customer is on an SKU, they have assurance of the same price for many years.
• Newer SKU pricing: Newer generation SKUs may have different pricing due to underlying hardware cost or additional features, but this does not affect existing customer plans.

Tarun Dua: "Typically, we do not change pricing of majority of our SKUs unless there is an (Inaudible) mismatch... once the customer is on an SKU, they have the assurance that they will have the same price for many years to come."

Product & Platform Development

• Tir AI/ML platform: New features include inference capabilities for standardized endpoint deployment of models in a scalable fashion, dataset features with quick-mount options for training or inference, and a committed node plan allowing customers to cut costs by prepaying for longer periods.
• Container registry: A new E2E Container Registry was launched to support Cloud native, Kubernetes, and Docker container features, with the ability for customers to clone existing registries.
• Security and compliance: New monitoring capabilities were added for early modeling of potential intrusion telltale signs, vulnerabilities, and compliance monitoring of nodes.
• Scalable file system: Added the ability to snapshot, resize, clone, and backup volumes on the scalable file system.

Margin Profile & Trajectory

• Q2 PAT margin: The 26.85% PAT margin for Q2 and 31.14% for H1 benefited from a change in depreciation method from WDV to SLM effective June 2023, per Schedule 2 of the Companies Act; the impact was ~INR 3 Cr for Q2 and ~INR 5.6 Cr for H1.
• Long-term margin view: Over the longer term, management does not see EBITDA or PAT going down; as more value is built into software, newer SKUs with a higher value proposition would have a significantly better margin profile because of a bigger software capability component.
• Asset turn trajectory: Management wants asset turn to have an upward trajectory, but it will not change quarter-on-quarter much; the long-term view is that as more software features add value, asset turn would increase.
• Payback lens: Management does not calculate asset turn as a primary metric; what they calculate is a payback period of 12 to 20 months across SKUs.

Tarun Dua: "Over a longer term I don't see any kind of going down on either EBITDA or the PAT line. I think over a longer term we should be able to maintain those."

Demand Environment

• AI/ML demand drivers: Demand for AI/ML workloads is driven by various sectors including higher education and research, with an impetus to localize data due to sectoral regulations and data protection requirements.
• Industry tailwind: India is expected to see data center capacity grow by over 10X over the next 5 years or so, per industry reports cited by management.
• Sales cycle reality: From deployment to significant uptake is about a 6-month cycle; the entire uptake can be anywhere between 5 to 12 months, meaning capacity deployed today converts to revenue with a lag.

Tarun Dua: "Overall we've always picked up something like a 6 months cycle from the day you have put up a deployment to where you see the significant uptake from that deployment."

Guidance & Commitments

Key Metrics