Neysa Networks booked no operating revenue at all in the financial year it raised its first big cheque. That same young company, still short of its third birthday, was valued at roughly $1.4 billion by Blackstone in February 2026.
The gap between those two facts is the whole story. Neysa sells the GPU infrastructure that Indian banks, government bodies and AI startups need to train and run artificial intelligence models without shipping their data abroad, and its rise from a zero-revenue filing to a Blackstone-anchored unicorn in under two years says as much about India’s AI-compute shortage as it does about the two data-centre veterans who built it.
Quick facts
| Company | Neysa Networks Private Limited (“Neysa”) |
| Founded | 2023, Mumbai |
| Founder(s) | Sharad Sanghi (co-founder and CEO) and Anindya Das (co-founder) |
| Businesses | Velocis (GPU and AI acceleration cloud), Overwatch (AI-native network observability), Aegis (AI security) |
| Latest FY revenue | No operating revenue reported for FY24 (year to March 2024); only interest income of Rs 13.87 lakh, as per Entrackr’s Fintrackr |
| Latest FY profit/loss | Net loss of Rs 3.1 crore in FY24, as per Entrackr’s Fintrackr |
| Listed | Private; no IPO. Reports say an eventual listing is under consideration, unconfirmed by the company |
| Market value / last valuation | About $1.4 billion as of the February 2026 Series B, as reported by TechCrunch and Entrackr |
| Key shareholders / CEO | CEO Sharad Sanghi; Blackstone holds a majority stake post-Series B; other investors include Nexus Venture Partners, Z47 and NTT Venture Capital |
What they do
Neysa builds and rents out AI infrastructure. Its core product, Velocis, gives enterprises, research institutes and government agencies on-demand or dedicated access to GPU clusters, mainly Nvidia H100 and H200 chips, along with an MLOps layer for training, fine-tuning and deploying models. Two supporting products round out the platform: Overwatch, which does AI-driven monitoring of networks and infrastructure, and Aegis, which secures AI systems against threats such as model and data poisoning. The pitch to customers is straightforward: keep sensitive data inside India, get GPU capacity without competing with global hyperscalers for it, and get a level of hands-on support that Sanghi says the giants will not offer a mid-sized Indian enterprise.
The origin
Sharad Sanghi had done this before. In 1998 he founded Netmagic Solutions on a simple observation: Indian businesses moving onto the internet needed dedicated data centres, not spare server space bolted onto someone else’s operation. He ran Netmagic for more than two decades, took it through three funding rounds backed by Nexus Venture Partners, Fidelity, Cisco and Nokia Growth Partners, and led its 2012 majority buyout by NTT Communications of Japan, staying on to run NTT’s data-centre and cloud business in India.
The second insight arrived the same way the first one did: through customers asking for something that did not exist yet. In early 2023, while still running NTT’s global data-centre operations, Sanghi found enterprise clients asking whether NTT could also provide GPU infrastructure for the AI workloads they were suddenly building, a few months after ChatGPT’s late-2022 debut had made generative AI board-level urgent. He resigned from NTT in June 2023, at 56, to start over rather than retire, and brought in Anindya Das, a colleague from the Netmagic days who had gone on to build cloud and network operations at NTT, as co-founder. Their working relationship stretched back roughly two decades before Neysa’s first line of code.
The struggle years
The company’s first fiscal year on paper was unglamorous. For the year ended March 2024, Neysa’s regulatory filings, as analysed by Entrackr’s Fintrackr in October 2024, show no operating revenue at all: the only income was Rs 13.87 lakh in interest on bank deposits, against a net loss of Rs 3.1 crore. That is despite having already closed a $20 million seed round in April 2024 led by investors including Z47, Nexus Venture Partners and NTT Venture Capital. The company was, in effect, spending seed money to build a product it had not yet sold.
The second, less visible struggle was scale. When TechCrunch profiled Neysa around its Series A close in October 2024, the company had 55 employees, about 12 paying customers and six large proof-of-concepts, and had only launched its flagship Velocis platform that July. It was trying to win enterprise AI workloads away from AWS, Google Cloud and Microsoft Azure, and compete for the same GPU allocations as far better-capitalised global neoclouds such as CoreWeave, at a moment when India itself had fewer than 60,000 deployed GPUs in total, a fraction of what the US or China had racked up, as TechCrunch reported in February 2026. Every GPU Neysa wanted to deploy was a GPU it had to fight the rest of the world for.
The turning point
The turning point has a precise date: 16 February 2026, when Blackstone announced it would lead a financing of up to $1.2 billion into Neysa, split between roughly $600 million of primary equity, giving Blackstone a majority stake, and up to $600 million of debt to fund GPU purchases, according to TechCrunch and Entrackr. Before that day, Neysa was a $128 million company on paper: that was its post-money valuation after the $30 million Series A it closed in October 2024, per Entrackr’s Fintrackr analysis of the round. After it, Neysa was priced at roughly $1.4 billion, an increase of about eleven times in sixteen months, and had gone from around 1,200 live GPUs to a mandate to deploy more than 20,000. Entrackr and other outlets described it as India’s first AI-focused unicorn of 2026, and its second unicorn overall that year after fintech firm Juspay.
The money behind it
- Seed, April 2024: $20 million, among India’s larger seed rounds that year, with backers including Z47 (formerly Matrix Partners India), Nexus Venture Partners and NTT Venture Capital, as reported by Business Standard.
- Series A, October 2024: $30 million, co-led by NTT Venture Capital (which put in about $8.9 million), Z47 and Nexus Venture Partners, with Anchorage Capital and angel investor Rajesh Kumar Dugar also participating; post-money valuation about $128 million, as per Entrackr’s Fintrackr.
- Series B, February 2026: up to $1.2 billion, split into roughly $600 million of primary equity led by Blackstone (taking a majority stake) and up to $600 million of debt financing for GPU procurement; co-investors included Teachers’ Venture Growth, TVS Capital, 360 ONE Asset and repeat backer Nexus Venture Partners. Valuation: about $1.4 billion, per TechCrunch and Entrackr.
- Total raised to date: $50 million in venture equity before the Series B, plus up to $1.2 billion in the Series B round, per TechCrunch.
What each backer changed: Nexus Venture Partners is the only investor in all three rounds, a continuity signal that predates Neysa itself, since Nexus also backed Sanghi’s Netmagic. NTT Venture Capital’s participation, coming from the same corporate family that acquired Netmagic in 2012, gave Neysa an early credibility marker inside the data-centre industry. Z47 and Nexus together held the two largest external stakes after the Series A, at 16.22% each, per Entrackr’s Fintrackr. Blackstone’s entry in 2026 changed the company’s category entirely, moving it from venture-funded software company to a private-equity-backed infrastructure buildout with a dedicated debt facility for hardware.
How it makes money
Neysa earns money by renting out GPU capacity and the software layered on top of it, rather than by charging a pure software subscription. Money comes in through three broad lines:
- GPU cloud compute (Velocis): customers pay for on-demand or reserved access to H100 and H200 GPU clusters, either in shared public-cloud form or as dedicated private clusters; as of October 2024, about 70% of deployed capacity sat in private clusters and 30% in public cloud, per TechCrunch.
- Platform and MLOps services: consulting and tooling to help enterprises right-size infrastructure, fine-tune models and manage the machine-learning lifecycle, sold alongside the raw compute, per TechCrunch’s October 2024 reporting.
- Observability and security add-ons: Overwatch (network and infrastructure monitoring) and Aegis (AI security, guarding against model and data poisoning) are sold as adjacent products, including through a September 2026 partnership with telecom software firm Mavenir to bring AI-native infrastructure to telecom operators, per Business Standard and GlobeNewswire.
Money goes out mainly on GPUs themselves: by April 2025 Neysa had put more than $42 million of its raised capital into cloud infrastructure capex, according to founder comments on the Neon Show podcast. That capital intensity is the part outsiders tend to underrate: unlike a typical software company, Neysa’s gross margin depends heavily on keeping expensive, depreciating hardware highly utilised, which is why the Series B was structured with a dedicated debt tranche for GPU purchases rather than funded entirely from equity. Sanghi has framed the company’s edge over AWS, AZURE and Google Cloud as service rather than price alone, pointing to round-the-clock support with what he has described as a 15-minute response time, a company-stated figure rather than an independently audited one.
The numbers
Neysa is genuinely young: it was incorporated during the financial year ended March 2024, so only one full year of audited financial data has been independently reported so far. Figures below are in Rs crore unless noted; where a year’s data has not been confirmed through a primary filing, it is marked as such rather than estimated.
| Fiscal year | Revenue (Rs crore) | Net profit / (loss) (Rs crore) |
| FY23 (to March 2023) | Not applicable — company not yet incorporated | Not applicable |
| FY24 (to March 2024) | Nil operating revenue; Rs 0.14 crore interest income | (3.1) |
| FY25 (to March 2025) | Not independently verified as of this research; third-party estimates conflict and are not used here | Not independently verified |
What is independently documented instead is the pace of operational scale-up:
- October 2024: 55 employees, about 12 paying customers and 6 large proof-of-concepts, per TechCrunch.
- April 2025: 65 employees, roughly 1,200 GPUs deployed and about 15 enterprise clients, per Sanghi’s comments on the Neon Show podcast.
- August 2025: headcount around 97, per Tracxn’s company profile.
- February 2026: 110 employees across Mumbai, Bengaluru and Chennai, and about 1,200 GPUs live at the time of the Series B announcement, with a mandate to deploy more than 20,000, per TechCrunch.
Read together, the trajectory is a company whose headcount and infrastructure footprint grew several times over in under eighteen months, while its only independently confirmed annual revenue figure remains zero. That is not unusual for infrastructure-heavy startups pre-scale, but it does mean the FY24 filing and the February 2026 valuation sit at very different points on the same company’s growth curve.
Where the money comes from
- By deployment type: about 70% of Neysa’s capacity was running as dedicated private clusters and 30% as shared public cloud, as of October 2024, per TechCrunch — a mix skewed toward the higher-touch, higher-price end of GPU hosting rather than commodity public-cloud rental.
- By customer sector: research institutes, AI-native startups, banking and financial services, manufacturing and media, per TechCrunch’s October 2024 reporting; by 2025 the customer and pilot base had grown to more than 20 across financial services, healthcare and media, per Forbes India.
- By geography: currently India-only. International expansion is planned, with Sanghi telling Forbes India in 2025 that global expansion was targeted within 12 to 18 months, and telling the Neon Show podcast in April 2025 that overseas deployment was planned for the “next fiscal year.”
The surprise, relative to how most people picture a “cloud” business, is that Neysa leans toward dedicated, contractually locked-in private clusters for regulated customers such as banks and government bodies rather than the flexible, pay-by-the-minute public cloud usage that dominates at global hyperscalers. That is a direct product of its sovereign-data pitch: customers in banking, healthcare and government are paying, in effect, for data residency and control as much as for raw compute.
The risks
- Debt-funded, depreciating hardware: roughly half of the up-to-$1.2-billion Series B is structured as debt earmarked for GPU purchases rather than equity, per TechCrunch’s February 2026 reporting. GPUs lose value and get outpaced by newer chip generations faster than typical infrastructure debt amortises, so a slowdown in customer demand or utilisation would leave debt service obligations sitting on top of fast-depreciating collateral.
- Sector-wide hardware cost volatility: GPU cloud capacity pricing has been rising across the industry; rival neocloud Nebius raised on-demand capacity prices and its preemptible-capacity prices further in 2026, according to HostingJournalist.com and Global Banking & Finance Review. Because Neysa’s own margin depends on the spread between what it pays for GPUs and what it charges customers on multi-year contracts, industry-wide hardware cost swings of this kind squeeze resellers who have already locked in customer pricing.
- Thin, concentrated customer base against fixed infrastructure costs: Neysa had about 12 paying customers and 6 large proof-of-concepts as of October 2024, growing to roughly 15 enterprise clients by April 2025, per TechCrunch and the Neon Show podcast. Fixed, capital-intensive GPU infrastructure needs high utilisation to make sense financially; a customer base this small means the loss or delay of even one or two anchor contracts could disproportionately hurt utilisation and revenue.
The takeaway
Neysa’s story is less about artificial intelligence than about timing a second act around a shortage you already understand. Sanghi did not build a data-centre company at 32 and an AI-cloud company at 56 because he had a new insight about neural networks; he built both because he recognised, twice, that Indian businesses were about to need infrastructure that did not exist domestically yet, and that being early with the right relationships (with investors, with hyperscaler-adjacent partners such as NTT, with the physical facilities of his own former company) mattered more than being first with a novel idea. The lesson that travels beyond AI cloud is about sequencing: raise conviction capital before the revenue exists to justify it, spend the early quiet year building the plumbing nobody sees, and be ready to convert scarcity in a market into a scaled, funded business the moment institutional capital decides the scarcity is real. Whether Neysa’s underlying unit economics catch up to its $1.4 billion price tag is a separate question from whether the strategy of getting there was sound, and it is a question the company’s next set of filings, not this one, will have to answer.
Frequently asked questions
What does Neysa Networks do?
Neysa provides GPU-based AI infrastructure and platform services under its Velocis brand, plus AI network monitoring (Overwatch) and AI security (Aegis), mainly for Indian enterprises, research institutes and government bodies that want to train and deploy AI models without using global hyperscalers.
Who founded Neysa and when?
Neysa was founded in 2023 by Sharad Sanghi, previously the founder and long-time CEO of data-centre company Netmagic (acquired by NTT in 2012), and Anindya Das, a former colleague from Netmagic and NTT who later built cloud and network operations there.
How much funding has Neysa raised and at what valuation?
Neysa raised $20 million in seed funding in April 2024 and $30 million in a Series A in October 2024 at a post-money valuation of about $128 million, per Entrackr. In February 2026, Blackstone led a financing of up to $1.2 billion (about $600 million in equity plus up to $600 million in debt), valuing Neysa at roughly $1.4 billion, per TechCrunch and Entrackr.
Is Neysa profitable?
No. The only independently reported annual financial filing, for the year to March 2024, showed no operating revenue and a net loss of Rs 3.1 crore, per Entrackr’s Fintrackr. More recent annual figures have not been independently verified as of this article.
Who are Neysa’s main competitors?
Global hyperscalers AWS, Microsoft Azure and Google Cloud, plus specialised GPU-cloud “neoclouds” such as CoreWeave and Lambda Labs, per TechCrunch’s October 2024 reporting on the company.
Sources
Figures are as of September 2026. Currency converted at $1 ≈ ₹96.0 as of 18 September 2026 (Trading Economics).
- TechCrunch, “Blackstone backs Neysa in up to $1.2B financing as India pushes to build domestic AI compute,” February 2026
- TechCrunch, “India’s Neysa bags $30M to compete with global AI hyperscalers,” October 2024
- Entrackr / Fintrackr, “Decoding Neysa’s Series A round, valuation and shareholding pattern,” October 2024
- Entrackr, “Gen AI startup Neysa turns unicorn after Blackstone-led $1.2 Bn funding,” February 2026
- Business Standard, “Sharad Sanghi’s AI cloud startup Neysa raises $20 mn in seed funding led by Matrix Partners,” April 2024
- Forbes India, “Anindya Das and Sharad Sanghi’s Neysa: Democratising AI adoption for enterprises,” AI Special 2025
- Founder Thesis, “The Infrastructure Prophet: How Sharad Sanghi Built India’s Digital Backbone Twice,” 2025/2026
- Neon Show / Neon Fund podcast, “0 to $1M in 6 months: Sharad Sanghi on why local AI clouds matter now,” April 2025
- Tracxn, Neysa Networks Private Limited company profile, accessed September 2026
- HostingJournalist.com, “Nebius Raises GPU Cloud Rates Again as AI Demand Accelerates,” 2026
- Global Banking & Finance Review, “Nebius Raises AI Cloud Prices Again Amid Surge in Computing Demand,” 2026
- Business Standard / GlobeNewswire, “Mavenir and Neysa Partner to Bring AI-Native Infrastructure to Operators, Enterprises, and Neoclouds,” September 2026
Found an error? Write to us and we’ll correct it in the open, dated, on the piece.