E2E Networks Q4 FY26: GPU scale-up drives operating leverage, while depreciation still weighs on full-year profit

E2E Networks Limited positions itself as an AI-first cloud GPU platform, offering cloud infrastructure and GPU capacity alongside its in-house AI/ML platform, TIR. In Q4 FY26, the operating story strengthened sharply. Operational revenue rose to INR 956 million, up 185.4% year-on-year and 36.6% quarter-on-quarter. EBITDA reached INR 581 million and EBITDA margin expanded to 60.7%. Profit after tax was INR 64 million, reversing the prior quarter loss.

The full-year picture looks very different, largely because of accounting effects from aggressive infrastructure investments. For FY26, operational revenue was INR 2,456 million and EBITDA was INR 1,263 million. Yet reported PAT was a loss of INR 156 million, which management attributed entirely to depreciation from GPU infrastructure capex.

What powered Q4: utilization, operating leverage, and monetizing larger clusters

Management described Q4 as evidence that the company can scale AI infrastructure rapidly, achieve high utilization, and convert capacity into revenue growth. On the call, the CFO highlighted a sequential swing in profitability, with PBT turning positive at INR 86 million in Q4 compared to a loss in Q3.

A key operating datapoint was utilization. Management stated that overall infrastructure utilization was 80% plus in March, and less than 85%, indicating both strong demand and remaining headroom. They clarified that this utilization measure covers the entire infrastructure base, including CPUs, GPUs, and storage.

The company also stressed progress on its TIR platform, stating that it supported large-scale GPU clusters and helped LLM teams run training on larger deployments. The commentary emphasized in-house capability across multiple layers of the stack and ongoing software improvements to reliability, performance, and scalability.

Financial snapshot (as disclosed)

The core tension in FY26: operating strength versus depreciation load

E2E’s FY26 income statement shows the trade-off typical of a rapid capacity build phase. EBITDA grew, but depreciation rose far faster because the company has been deploying GPU infrastructure at scale.

Management explicitly said the business is “strongly cash positive” at the EBITDA level and that reported profitability should improve as utilization continues to ramp and revenue outpaces depreciation. The CFO also clarified that GPU infrastructure is depreciated over six years.

On the balance sheet, property, plant and equipment increased to INR 14,966 million as of March 2026 from INR 9,471 million in March 2025. The company also reported CWIP in its capex disclosures: FY26 capex of INR 6,962 million includes INR 5,334 million of CWIP for GPUs under deployment.

Capacity expansion and funding: Blackwell rollout and multiple financing paths

Capacity expansion remains central to the investment narrative. The investor presentation highlights GPU scale reaching around 3,900 in FY25 and around 5,050 cloud GPUs in the current footprint description. It also outlines Blackwell expansion: 1,024 B200 GPUs and 128 RTX 6000 Pro, with another 1,024 B200 stated as in process.

On the call, management guided that the first 1,024 B200 cluster is targeted to go live before mid-May, while acknowledging supply-chain delays. They also said another 1,024 B200 cluster has already been planned for deployment a couple of months later.

Funding disclosures were detailed in the presentation. The company reported preferential issue fund raises totaling INR 14,849.34 million across Q2 FY25, Q3 FY25, and Q4 FY26, with utilization across FY25 and FY26 and a remaining balance of INR 2,314.70 million.

E2E also indicated that it is expanding beyond the traditional equity-versus-debt approach. Management said it is exploring private credit and asset-light models to accelerate GPU expansion. They discussed an MoU with L&T to monetize GPU infrastructure being built by L&T, while clarifying that the arrangement is exploratory and not exclusive.

Demand, pricing, and customer mix: strong backdrop, limited granularity

Management characterized the demand environment as strong both in India and globally, and noted that the company prioritizes India-first workloads while also serving international demand. A data point from the call indicated that international revenue in Q4 was roughly 35% to 37%.

The company also said that government business for the quarter did not exceed 35% to 40%. While the presentation references a contract under the India AI Mission, management emphasized that it does not force offtake under the mission and will sell unutilized capacity to other customers.

On pricing, management stated it is not seeing negative pressure and suggested tailwinds that are “inching up the prices,” while also describing the need to balance short-term realization with longer-term customer relationships.

However, the call also revealed limits in disclosures. Management repeatedly declined to provide detailed MRR splits by India AI Mission versus enterprise, GPU versus CPU contribution to MRR, and contract-duration mix, citing rapid week-to-week variability and the still-small base of installed GPUs.

Takeaways from Q4 FY26

E2E Networks delivered a strong Q4 FY26 operational performance, with sharp revenue growth and a high EBITDA margin, indicating operating leverage as utilization improves. At the same time, FY26 reported profitability remains constrained by depreciation from the capex-heavy GPU build cycle.

The near-term execution milestones are clear. The company has guided a mid-May target for the first 1,024 B200 Blackwell cluster going live, with another 1,024 planned soon after. If deployment timelines hold and utilization remains strong, the gap between EBITDA performance and reported PAT could narrow over time. But investors will still need to track the pace of depreciation, utilization outcomes, and the economics of any asset-light structures once finalized.