Q2 FY27 · NSE: E2E

E2E Networks

Results filed 16 October 2026 · Market cap ₹10,698 crore · PE 91.2

TurnaroundTranscript read

Revenue: +410.9%; on a year ago
EBITDA: +852.6%; on a year ago
Profit to shareholders: Back to profit; from a loss a year ago
EBITDA margin: 76.6%; +1.4 pts on last quarter

E2E Networks reported strong Q2 FY27 performance, with operational revenue rising 410.8% YoY to ₹2,238 Mn and PAT reaching ₹727 Mn alongside a 76.6% EBITDA margin. The rapid scale-up was anchored by the full deployment and multi-year contracting of 1,024 NVIDIA Blackwell GPUs under a ~₹1,000 Cr sovereign agreement. However, balance-sheet borrowings jumped more than fourfold during the half-year to ₹4,504 Mn, while management declined to share operational KPIs such as fleet utilization or contracted revenue mix during the earnings call.

E2E Networks Limited, Q2 FY27₹ crore, consolidated

	Q2 FY27	Q1 FY27	vs previous quarter	Q2 FY26	vs a year ago
Revenue	224	157	+42.7%	43.8	+410.9%
Operating profit (EBIT)	97.49	57.26	+70.3%	−24.8	n.m.
Operating margin	43.6%	36.5%	+704 bps	-56.6%	+10019 bps
Other income	9.93	11.42	−13.0%	8.3	+19.6%
Exceptional items	0	0	–	0	–
Profit before exceptional items and tax	97.13	58.63	+65.7%	−18.49	n.m.
Profit to shareholders	72.7	43.88	+65.7%	−13.46	n.m.
Earnings per share (₹)	3.54	2.14	+65.4%	-6.75	n.m.

Source: E2E Networks Limited’s filing to NSE, 16 October 2026. Operating profit is revenue less every expense except interest, so it leaves out other income and exceptional items. n.m.: not meaningful, because the earlier figure was zero or negative, or the latest one is a loss, or, for earnings per share, because the number of shares changed.

What stood out

Signed a ~₹1,000 Cr multi-year Sovereign AI contract for Blackwell GPUs running through June 2029
Source
Press Release page 2: 'Entered into a binding term sheet of ~₹1,000 Cr (excl. taxes) with an Indian sovereign AI company for NVIDIA Blackwell cloud GPUs and allied services, valid through June 2029.'
Operational revenue grew over 5.1x YoY and EBITDA margin expanded to 76.6%
Source
Press Release page 2 / Presentation page 19: 'Revenue of ₹2,238 Mn, up 410.8% YoY and 42.7% QoQ; EBITDA of ₹1,715 Mn at a 76.6% margin (+141 bps QoQ; +3,553 bps YoY)'
Rapid build-up of debt to fund expansion
Source
Presentation page 21: Balance sheet borrowings rose from ₹1,032 Mn as of March 31, 2026 to ₹4,504 Mn as of September 30, 2026.

Guidance

Metric	Guidance	Period
EBITDA Margin Source Concall page 4 (Tarun Dua: 'See, overall, we are looking at this as a 70% plus EBITDA margin business. That's what we feel for the medium term and the long term.')	70% plus EBITDA margin expected in the medium to long term	Medium to long term
GPU Deployment Source Concall page 13 (Tarun Dua: 'This quarter, we expect the deployment to go through for the 1,024.')	1,024 B200 GPUs planned to go live during Q3 FY27	Q3 FY27

Management repeatedly emphasized their policy of not giving forward-looking revenue, MRR, or capex guidance ('We don't give any guidance. We always say, look back at what we have done rather than giving a future guidance' - Concall p. 4).

Tailwinds

Full operational contribution and utilization of NVIDIA Blackwell B200 GPUs
Source
Concall page 3 & 13 (Tarun Dua: 'the flagship Blackwell capacity is fully utilized in the particular contract that was mentioned.')
Expansion of EBITDA margins to 76.6% via operating leverage and higher average realizations
Source
Concall page 3 (Nitin Jain: 'The performance is driven by contribution from the Blackwell GPU, better utilization, and higher average realization across the GPUs.')
Sequentially higher exit MRR
Source
Presentation page 18: Month Exit Revenue (MRR) expanded from ₹718 Mn in June 2026 to ₹751 Mn in September 2026.

Headwinds

Surge in financial leverage and interest costs
Source
Presentation page 19 & 21: Finance costs grew 415.6% YoY to ₹103 Mn in Q2 FY27 as borrowings rose to ₹4,504 Mn.
Substantial step-up in depreciation expense from asset capitalization
Source
Presentation page 19: Depreciation rose 72.8% YoY and 22.0% QoQ to ₹740 Mn for the quarter.
No advance customer contracts finalized yet for upcoming B200 GPU lot
Source
Concall page 17 (Tarun Dua: 'No, we have not decided anything around the next lot of B200.')

New businesses and products

capacity expansion · In deployment; expected to go live in Q3 FY27

Deployment of second tranche of Blackwell B200 GPUs (1,024+ GPUs)

Expected benefit: Expands GPU fleet from ~5,100 to over 6,100 units to meet compute demand
Risks: Procurement and delivery cycles typically taking 3 to 6 months; uncertainty over vendor advance payments
Timeline: Q3 FY27

Source

Concall page 13 (Tarun Dua: 'This quarter, we expect the deployment to go through for the 1,024.') and Presentation page 11 ('IN DEPLOYMENT: Blackwell scale-up 1024+ B200s planned')

new product · Launched / added

VPC Connect on TIR AI/ML Platform

Expected benefit: Delivers secure, private networking and enterprise-grade isolation across the AI lifecycle for regulated workloads at par with global hyperscalers
Timeline: Q2 FY27

Source

Press Release page 2: 'TIR AI/ML Platform: Added VPC Connect enables secure, private networking for GPU instances, training clusters and inference endpoints...'

new product · Soft release; official launch a few weeks away

Jarvislabs.ai Managed Endpoints, Model APIs and Serverless Inference

Expected benefit: Production-ready open-weight models, pay-per-token APIs for open-source models, and InfiniBand H200 cluster support
Risks: Early stage of customer adoption
Timeline: Official launch in a few weeks from mid-October 2026

Source

Concall page 10 (Tarun Dua: 'I wouldn't say launched as much as basically release, like soft release. Obviously, the official launch might be a few weeks or so away.')

Things to check

Revenue up over 60%: check for an acquisition or a weak year-ago quarter
Significant increase in leverage with no stated debt limit
Source
Borrowings escalated from ₹1,032 Mn in Mar-26 to ₹4,504 Mn in Sep-26 (Presentation p.21), and the MD stated: 'we don't have an upper limit defined for ourselves that, okay, we can't go beyond this on debt' (Concall p.7).
Customer/contract concentration with limited operational disclosures
Source
A single Sovereign AI contract of ~₹1,000 Cr ties up the entire 1,024 B200 fleet (Concall p.13), but management declined to specify how much MRR remains to be recognized from it or provide customer concentration metrics (Concall p.4).

How the call went

The call lacked operational detail. While top-level financials were shared, management deflected inquiries regarding capacity utilization, customer/contract revenue split, debt tolerance, and procurement advance terms behind macro commentary.

Specific or vague: Low; management refused to provide hard numbers on utilization rates, workload breakdown (inference vs training), contracted vs spot revenue percentages, or specific capex quantum, arguing that point-in-time metrics fluctuate excessively.
Candour: Low; the MD deflected numerous detailed analytical questions by invoking broad market generalities like a $1 trillion TAM and stating the company does not predict the future.
The hard questions: Analysts probed on key vulnerabilities: whether the ₹1,000 Cr contract was already fully baked into MRR, sustainable gross margin limits, comfortable debt thresholds, whether GPU suppliers required advances, and competition from large Indian peers (e.g. L&T scaling 10,000 B300s or others with 90,000 GPUs). Management mostly declined to give precise answers.

What analysts pressed on

Concentration and ramp-up trajectory of the ₹1,000 Cr sovereign contract within MRR.
The ceiling for balance sheet debt following a steep jump in borrowings and an enabling ₹1,500 Cr fundraise resolution.
Competitive positioning and differentiation against large-scale entrants deploying 8,000 to 90,000 GPUs in India.
GPU procurement lead times, vendor advance terms, and NVIDIA allocation backlogs.

Answers that sidestepped

On whether MRR is fully baked in from the ₹1,000 Cr contract: 'See, I don't know how to answer that question. You are asking a question about a future that I cannot predict.' (Concall page 4)
On acceptable debt levels: 'We are very comfortable with debt as long as we have the ability to repay. So, we don't have an upper limit defined for ourselves that, okay, we can't go beyond this on debt...' (Concall page 7)
On inference vs training revenue mix: 'See, we look at the whole set of services as an overall whole, rather than looking functionally at, oh, revenue is coming from inference or training... So, any percentage we quote, we want to make sure that, that is something which is actually trackable and which is not a very ephemeral point-in-time number.' (Concall page 5)
On contracted revenue share: 'currently our size is too small for sharing point-in-time information, and these numbers would change dramatically.' (Concall page 8)
On whether advances must be paid for GPU orders currently: 'I am not sure how to answer that.' (Concall page 16)

Numbers against the story

The reported numbers show spectacular revenue growth (+410.8% YoY to ₹2,238 Mn) and high EBITDA margins (76.6%), validating management's narrative of rapid demand for sovereign AI compute. However, the narrative of software differentiation and unconstrained market opportunity downplays significant balance-sheet leverage, as borrowings grew by over ₹3,470 Mn in six months. Additionally, management's refusal to provide utilization rates, contracted revenue mix, or GPU advance details leaves key financial sustainability questions unanswered.

Sources

Concall transcript · filed 21-Oct-2026
Press release · filed 16-Oct-2026
Investor presentation · filed 16-Oct-2026
Figures: the company’s results filing with NSE, linked under the table.

Notes written by gemini-3.8-flash from the documents above on 1 November 2026. They summarise what the company said and can contain mistakes: check the source passage before relying on a point. Not investment advice.