Caught In The GPU Pricing Paradox, Can ESDS Deliver A Bumper IPO?

Two decades on, enterprise cloud and AI ecosystem provider ESDS Software Solution is setting sail for its next big quest — public markets. The Nashik-based company’s IPO, which comprises an entirely fresh issue, opens tomorrow. The price band has been fixed at ₹408 to ₹429 per share. 

The ₹720 Cr ESDS IPO comes as India’s data-centre industry shifts from traditional cloud setup to AI infrastructure. The cloud services provider plans to use the IPO proceeds to fund a major expansion of data-centre infrastructure, including GPUs and servers, as ESDS looks to capture the growing demand for AI compute and sovereign cloud.

However, there is a catch. The infrastructure ESDS needs has become significantly more expensive, while competition for GPUs and AI workloads is intensifying. 

Servers that were quoted at around ₹25 Lakh when ESDS first filed its draft offer documents cost approximately ₹1.8 Cr by the time it prepared its red herring prospectus. Similarly, prices of GPUs, storage, firewalls and networking equipment have also increased since.

What this means is that the money set aside for the expansion will buy fewer assets than ESDS initially expected, even as the company looks to make GPU-as-a-service and AI compute its next growth engine.

So, what exactly are investors buying into with the ESDS IPO? Is it a profitable cloud and data-centre business with an established customer base, or a bet on its ability to become a meaningful player in India’s emerging AI infrastructure market? The answer lies as much in understanding how ESDS has evolved over the past two decades as in assessing whether the business it has built can give it an edge in its next phase of growth.

From Back-Office Support To Community Cloud

For ESDS, the IPO comes years after founder, chairman and managing director Piyush Somani began securing back-office work from data centres in the US and web-hosting companies in the UK.

Born to a banker, Somani turned to entrepreneurship at his father’s insistence, choosing to build a business rather than spend his career in a job. “He wanted me to become a job creator rather than a job seeker,” Somani said.

Somani registered ESDS in 2005 and used the earnings from its overseas business to pursue an opportunity in India’s data centre market. In 2007, he told his employees that the company would build a data centre in India. The plan was met with scepticism, but the facility eventually went live in 2009.

Around that time, the company began experimenting with autonomous scaling technologies. Soon after, it introduced vertical and diagonal auto-scaling technology in 2011, incorporating AI into its cloud infrastructure. The Indian cloud market was still nascent back then. 

This prompted ESDS to change its sales model. It began pitching cloud services to financial institutions such as banks on a pay-per-transaction, pay-per-branch and pay-per-user basis, eventually building a banking community cloud. The experience shaped ESDS’ early focus on sovereign cloud. 

“We brought it in the form of a community cloud,” Somani said, arguing that the concept has since become increasingly important globally. 

A community cloud is a cloud computing environment shared by multiple organisations that have common goals, security requirements, or regulatory needs. More than 170 banks currently use its banking community cloud, while over 100 government organisations have adopted its government community cloud.

The company later extended the model to SAP HANA and other government workloads. Its BFSI community-cloud business now covers core applications, digital payments, security, disaster recovery and managed infrastructure for banks, NBFCs and other financial institutions.

ESDS’ work with Software Technology Parks of India, or STPI, added another layer to this expansion from 2020. Three of its data centres operate from space leased from STPI, to which ESDS pays 18% of the revenue generated by these facilities annually.

In September 2021, ESDS made its first attempt to enter the public markets. It, however, did not proceed with the issue due to the pandemic. Somani said the experience changed how he viewed IPO.

“Never go to the market to raise money. Go to the market to get listed because you are already making money,” he said.

ESDS returned with another DRHP in March 2025, proposing to raise ₹600 Cr. SEBI approved the document in December 2025. The issue has since been increased to ₹720 Cr, with the company linking the larger capital requirement to growing demand for AI infrastructure.

What Has ESDS Built So Far?

ESDS entering the public markets is far broader than the back-office support business Somani started, with a diversified portfolio spanning cloud infrastructure, managed services and SaaS.

Its operations are divided into three main segments: infrastructure-as-a-service, managed services and SaaS. 

Its infrastructure-as-a-service business covers cloud and colocation services, including CPUs, RAM, GPUs, storage, firewalls and related infrastructure. Its managed services business covers databases, operating systems, cybersecurity, storage and network management, and its SaaS portfolio includes around 40 software services spanning application performance management, autonomous scaling and database administration.

The company reported revenue of ₹472 Cr in FY26, with profit after tax (PAT) at ₹120 Cr. According to its IPO presentation, ESDS had a revenue retention rate of 94.9% among existing customers, an EBITDA margin of around 50% and a PAT margin of roughly 25%.

It serves more than 2,500 customers, with government entities contributing a major part of the business. ESDS hosts platforms linked to NHAI, MHADA, the Public Distribution System, MUDRA Yojana, Stand-Up India, Udyami Mitra, PM SVANidhi, the Indian Ports Association, the Shipping Corporation of India and Maharashtra’s Revenue Department.

These customers give ESDS an established revenue base and a route into regulated industries where data location, security and compliance influence technology purchases.

The company is also expanding its physical footprint. A 20 MW data centre is being developed in Uttar Pradesh, while a 5 MW facility is planned in Kolkata. ESDS also plans to add another 5 MW of capacity at its Bengaluru campus. However, the opportunity it is pursuing is attracting global hyperscalers, domestic conglomerates and specialist GPU cloud providers.

India’s operational data centre capacity increased from 520 MW in 2020 to around 1.5 GW in 2025 and is projected to reach between 4.5 GW and 6.5 GW by 2030, according to a Council on Energy, Environment and Water study.

The report estimates that committed investments in India’s data centre ecosystem reached approximately $95 Bn between 2019 and 2025 and could exceed $100 Bn by 2027. The five largest operators accounted for around 66% of the country’s operational data centre capacity. 

Another layer is emerging on top of these facilities in the form of GPU-as-a-service. It allows companies to access GPU clusters through the cloud rather than purchase and maintain expensive equipment themselves.

ESDS sees this shift as an opportunity to move beyond conventional cloud colocation data centre services.

The company estimates that India’s traditional cloud services market is growing at around 30% annually, while the GPU infrastructure market is expanding at around 50%. Globally, it believes GPU-as-a-service is growing at a triple-digit rate. ESDS expects the segment to contribute a growing share of its future revenue.

Its GPU-as-a-service portfolio includes infrastructure based on NVIDIA and AMD chips for model training, inference, scientific computing and other AI workloads. ESDS is positioning itself as a provider of fully managed GPU clusters rather than merely a seller of computing capacity. 

But securing the latest GPUs remains difficult. Inc42 reported in July that the severe shortage seen during the early generative AI boom had eased, but global data centre demand continued to outpace supply. Access to newer, high-performance chips remained particularly constrained.

The demand-supply gap is one reason ESDS is paying considerably more for infrastructure than it expected when it began the latest IPO process, as highlighted earlier.

What Gives ESDS An Edge In The AI Race?

ESDS believes it can partly offset the capital intensity through customer advances. According to the company, AI infrastructure customers are increasingly willing to pay up to 12 months in advance before a cluster is delivered. This can cover roughly 30-35% of the capex requirement, with the remainder potentially financed through lenders or infrastructure partners.

This also points to one of the biggest constraints for India’s AI infrastructure boom: financing. Towards this, ESDS has signed a $1.25 Bn agreement with Australia’s Sharon AI that will own and operate an 8,000-GPU cluster, which ESDS will then use for serving its domestic and global customer needs for compute under a seven-year contract. Somani said that the company will explore similar partnerships going forward. 

Alongside GPUs, ESDS is building its pitch around sovereign cloud infrastructure. Its proprietary Swaraj Cloud platform has more than 50 services, compared with fewer than 10 services on its earlier Enlight Cloud platform. 

The company describes Swaraj as an indigenous cloud orchestration platform designed to reduce dependence on multinational technology providers while giving customers greater control over their infrastructure and data.

ESDS says Swaraj currently has more than 50 services, compared with fewer than 10 services in its earlier eNlight Cloud platform. These capabilities have also helped improve the company’s operating leverage, as software costs have fallen while revenue has expanded.

The company’s pitch around indigenous infrastructure also comes against a growing concern among Indian enterprises and regulators around data sovereignty, foreign technology dependencies and geopolitical risks.

India has more than 150 data centres, but remains far behind the US, China and the combined EU and UK markets in the infrastructure available to support AI compute, according to Inc42’s analysis of India’s sovereign AI opportunity. 

ESDS estimates that international cloud providers currently command around 95% of the Indian market, leaving Indian providers with roughly 5%. It believes this imbalance could change as enterprises and government organisations become more conscious of where data resides, who controls the underlying infrastructure and how data is encrypted.

The company is also betting that its enterprise relationships will help it compete beyond the sovereign-cloud mandate. It cited customers that have moved away from hyperscaler platforms after costs or changing requirements made those platforms less viable over time.

The company entering the public markets is profitable, serves more than 2,500 customers and already operates across cloud, managed services and software. But its real test begins after the money is raised.

ESDS will have to purchase increasingly expensive infrastructure, secure the latest GPUs, bring clusters online, attract customers and keep that computing capacity in use. At the same time, it will need to ensure that its AI push does not weaken the economics of the existing business supporting the transition.

The post Caught In The GPU Pricing Paradox, Can ESDS Deliver A Bumper IPO? appeared first on Inc42 Media.