# 1. Financial Performance ## A. Key Figures * **Revenue:** **₹95.6 Cr** Q4 (+186% YoY / +37% QoQ) · **₹245.6 Cr** FY26 (+50% YoY) * **EBITDA:** **₹58.1 Cr** Q4 (60.7% Margin) · **₹126.3 Cr** FY26 (+30.6% YoY) * PAT: INR64 Mn Q4 (vs. Q3 loss) · -INR156 Mn FY26 (Full-year loss) * **MRR (March):** **₹37.4 Cr** (vs. **₹29 Cr** Jan/Feb average) * **Depreciation:** **₹51.3 Cr** Q4 ## B. Revenue & MRR * **Exponential Growth Trajectory:** Record quarterly performance driven by high infrastructure utilization and successful execution of a deep-tech validated strategy. * **MRR Momentum:** Significant monthly recurring revenue spike in March reflects peak utilization of current assets; management notes hardware can be billed at varying rates depending on use-case complexity. * **Geographic Mix:** International clients now contribute a substantial **35% to 37%** of total revenue, diversifying the client base. ## C. EBITDA & Margins * **Margin Expansion:** Robust sequential improvement in EBITDA margins (up **413 bps**) driven by operational leverage, despite increased investments in high-level engineering talent. * **Strategic Hiring:** Rising employee costs are a deliberate investment in "high-value token" capture, aimed at securing market opportunities 12–24 months out. ## D. Depreciation & PAT * **Profitability Turnaround:** Achieved a significant quarterly swing to positive PAT as revenue growth began to outpace the heavy depreciation load from GPU infrastructure. * **Asset Amortization:** Infrastructure assets, including upcoming clusters, follow a strict **6-year** depreciation schedule; management expects revenue to progressively outrun these non-cash charges. * **Future Depreciation Headwinds:** Deployment of the new B200 cluster in **mid-May** will increase depreciation, though management expects the impact to be lower than the **₹25 Cr** per quarter estimated by the street. ## E. Capital Allocation * **GPU Footprint Expansion:** Capital deployment remains focused on rapidly scaling the GPU cluster; new B200 units are currently sitting in **Capital Work-in-Progress (CWIP)** pending imminent deployment. * **Judicious Deployment:** While FY27 capex is not yet finalized, the focus remains on maintaining profitability across all new structural business initiatives. --- # 2. Capacity & Infrastructure ## A. Key Figures * **GPU Deployment:** **2,048** B200 units planned FY26 · **1,024** units live starting May * **Utilization Rate:** **80%+** March average (Range: **70%–85%**) * **Total Capacity:** **3,900** units across CPU/GPU/Storage * **Asset Lifespan:** **7 to 8 years** projected for GPU hardware ## B. GPU Deployment & Scaling * **Next-Gen Integration:** Significant capacity expansion underway with NVIDIA Blackwell B200 units; current utilization metrics notably exclude this incoming capacity. * **Strategic Rollout:** Initial deployment of over **1,000 units** in May marks the first phase of doubling the planned annual GPU footprint. ## C. Utilization Rates * **Optimized Efficiency:** Infrastructure reached high utilization thresholds in March, providing a balance between rapid scaling and the elasticity required for further growth. * **India AI Mission Support:** Successfully fulfilled scaling requirements for mission allottees while maintaining steady utilization without over-provisioning clients. * **Performance Volatility:** Reported utilization serves as a historical average, fluctuating based on specific hardware configurations and varying billing rates. ## D. Hardware Lifecycle & Pricing * **Asset Longevity:** Management anticipates a robust multi-year operational life for GPU assets, supporting long-term ROI despite the fast-paced nature of AI hardware. * **Market Stabilization:** Local pricing peaks driven by supply shortages have subsided, leading to a more stable GPU pricing environment. --- # 3. Business Model & Strategy ## A. Key Figures * **Projected Revenue Mix:** **85%-90%** of MRR from GPU contribution in upcoming quarters ## B. Revenue Mix & Strategic Pivot * **Technology-First Valuation:** Management is pivoting the narrative from asset monetization to a **technology business** model, de-emphasizing traditional asset turn metrics in favor of platform-driven growth. * **Dynamic Contract Structure:** Revenue stability is managed through a judicious mix of short, medium, and long-term contracts, though no fixed allocations are maintained due to the dynamic market environment. * **GPU Dominance:** Revenue is expected to become heavily concentrated in GPU services, shifting away from legacy cloud offerings as the core growth engine. ## C. Asset-Light Expansion & Financing * **Hybrid Growth Model:** The company is transitioning toward an asset-light strategy, utilizing **private credit**, **structured financing**, and **partnerships** to scale GPU capacity without heavy equity dilution. * **Platform Monetization:** Expansion involves leveraging the **TIR stack platform** to manage third-party GPUs, creating new revenue lines by bringing more hardware under management. ## D. Token Value Strategy * **High-ROI Workloads:** Strategy is shifting toward capturing "high-value tokens"—prioritizing complex, high-accuracy AI outcomes and problem-solving over low-margin, transactional tasks like basic search. * **Value-Based Pricing:** Management is targeting customers who prioritize ROI and business outcomes over per-hour GPU pricing, viewing AI-driven software as a decadal growth opportunity. ## E. Partnership Frameworks * **Strategic Alliances:** An exploratory, non-exclusive MOU with **L&T** aims to monetize their infrastructure; meanwhile, the company remains aligned with the **India AI Mission** for sovereign AI tailwinds. * **Hardware Roadmap:** Infrastructure is being positioned to support next-generation **NVIDIA Hopper and Blackwell** GPUs as cluster requirements expand. --- # 4. Demand & Pricing ## A. Key Figures * **Spot Rate Inflation:** **25% to 30%** increase in H100 spot rates (April) * **Vertical Mix:** **35%-40%** of total business from Government/India AI Mission * **Demand Multiplier:** **10x to 50x** projected increase in future utilization levels ## B. Market Demand * **Secular AI Tailwinds:** Management views India as a nascent "AI Factory," driven by a structural shift where AI has moved from niche interest to a core business capability. * **Sustained Utilization:** Despite a massive historical surge in demand, current visibility remains robust for existing GPU inventories, supported by long-term production workloads rather than short-term projects. * **Next-Gen Interest:** Significant customer engagement is reported for upcoming **Blackwell capacity**, with the company currently aligning hardware combinations to specific client problem-solving requirements. ## C. Pricing Trends * **Pricing Resilience:** Rapid hardware cycles (Rubin/Blackwell) have not pressured rental rates; global supply shortages continue to outstrip demand, creating favorable pricing tailwinds. * **Budgetary Shift:** Demand is increasingly funded via "business capability" budgets rather than traditional IT caps, leading to lower price sensitivity as customers focus on high-value token generation. * **Strategic Balancing:** While spot prices have seen **20% to 25%** spikes, the company is prioritizing long-term ecosystem sustainability over short-term margin maximization. ## D. Customer Commitments & Vertical Mix * **Contract Strategy:** Pursuing **2 to 3 year** commitments for large-scale users while intentionally retaining capacity to capture the rising value of AI token generation. * **Revenue Composition:** Training workloads currently account for the majority of revenue, though hardware clusters are increasingly used flexibly for both training and inference. * **Diversified Base:** Beyond the significant government contribution, growth is balanced across digital natives, BFSI, and SMEs to mitigate reliance on any single segment. --- # 5. Supply Chain & Technology ## A. Product Roadmap & Procurement * **Next-Gen GPU Integration:** Deployment of **NVIDIA Blackwell GPUs** is slated for **mid-May 2026**, targeting global demand across enterprise, BFSI, and education sectors. * **Multi-Generational Strategy:** Management is executing a tiered rollout, maintaining robust demand for **Hopper and Ampere** architectures while preparing for future **B300, GB300, and Vera Rubin** cycles. * **Demand Dynamics:** New GPU architectures follow a "funnel-style" growth pattern, where legacy hardware remains highly utilized even as cutting-edge clusters are introduced. ## B. Software & Platform Capabilities * **Full-Stack Optimization:** Enhanced software architecture and the **TIR stack** now support large-cluster training and inference for both bare-metal and container-based deployments. * **Infrastructure Reliability:** Recent software overhauls have focused on improving the scalability and performance of the GPU infrastructure to support AI data scientist pipelines. * **Monetization & Time-to-Market:** Increased infrastructure monetization is driven by in-house capabilities across **all five layers** of the tech stack, accelerating customer token generation and deployment speeds. --- # 6. Risks & Cloud Infrastructure ## A. Operational Risks & Infrastructure * **Deployment Headwinds:** Early procurement of Blackwell GPUs has been offset by **global supply chain impacts** and component shortages, delaying live integration. * **Hardware Longevity:** Management dismisses rapid obsolescence risks, citing high **test-debug cycle costs** and platform complexities that lock in hardware utility beyond the typical 2-3 year window. ## B. Revenue Dynamics & Scaling * **Volatility Mitigation:** Historical revenue lumpiness is moderating due to a secular shift in demand; further stabilization is contingent on scaling the GPU fleet to **tens of thousands** of units. * **Reporting Thresholds:** Detailed segmentation and client profiling remain withheld to avoid misleading data fluctuations until the GPU base reaches critical mass. ## C. Capital Strategy * **Funding Roadmap:** To sustain infrastructure expansion and growth momentum, the company is actively evaluating a mix of **equity and debt** financing options. --- # 7. Guidance & Outlook ## A. Key Figures * **GPU Capacity Target:** **6,000 units** minimum baseline by FY27 * **Near-Term Visibility:** **2,048 GPUs** currently in pipeline ## B. Capacity & Growth Strategy * **Scalable Infrastructure Baseline:** Management view the FY27 unit target as a conservative floor rather than a ceiling, though they declined to formalize higher upside projections at this stage. * **Demand-Driven Guidance Policy:** Specific MRR and annual guidance withheld to avoid capping internal potential amidst robust demand and a rapidly evolving market landscape. * **Strategic Horizon:** Strong visibility into the next **52 weeks** of scaling, supported by a focus on aligning capacity with current market trends. ## C. Future Deployments * **Next-Gen Hardware Integration:** Planned deployment of **NVIDIA Blackwell GPUs** in May 2026, contingent on the delivery of critical components. * **Phased Cluster Rollout:** Sequential deployment strategy established for the first two clusters, with the second phase following the initial mid-May launch by **two months**. * **Revenue Recognition:** Customer profiles and revenue timelines for new deployments remain confidential until final deal execution.