Test report DSG-5776 · Rev A · tested October 30, 2026
AI Datacenter InfrastructureDevice under test
Neysa plans up to 9,000 GPUs for AI cloud at STTGDC’s Noida site
Neysa will place up to 9,000 GPUs at STTGDC’s Noida 2 site for Velocis, targeting Q1 2027 as India’s sovereign-cloud ambitions meet a 400MW-plus operator footprint.
- Read
- 3 min
- Words
- 583
- Node
- 28nm
- Operator
- Elena Vasquez
Spec summary
- Phase one supports up to 9,000 GPUs and targets a go-live around Q1 2027.
- STT Noida 2 offers 22MW; Noida 1 offers 17MW; Noida 3 is planned at 28MW.
- Neysa’s April 2025 partnership targets a 400MW GPU cluster containing 25,000 GPUs.
- KKR and Singtel fully acquired STT GDC from ST Telemedia in February before its STTGDC rebrand.
- STTGDC India has more than 400MW of critical IT load across 30 facilities in 10 cities.

By Elena Vasquez
Neysa will deploy up to 9,000 GPUs at STTGDC’s STT Noida 2 facility, with the first phase designed for AI training and inference and expected to go live around Q1 2027.
The installation will supply GPU infrastructure for Neysa’s Velocis AI cloud platform. APAC operator STTGDC India announced the project this week at its Uttar Pradesh site, starting the first phase of a multi-year working relationship with the Indian neocloud.
The company linked the agreement to growing demand for resilient, scalable infrastructure capable of running increasingly complex, compute-intensive workloads. That need extends beyond raw accelerator access, according to Bimal Khandelwal, CEO India at STTGDC.
What does the Noida deployment change for customers?
Khandelwal said: “India’s AI ambitions will depend not only on access to compute, but on the digital infrastructure required to support increasingly complex and compute-intensive workloads at scale.”
He described the project as a combination of Neysa’s specialized AI capabilities and STTGDC’s infrastructure expertise. Khandelwal said that pairing should help enterprises move from experimentation to real-world applications and support the ecosystem cooperation needed to accelerate AI adoption.
Khandelwal said he was proud to work with Neysa and its co-founder and CEO, Sharad Sanghi. Sanghi’s work building one of India’s earliest Internet data-center companies helped shape the country’s data-center industry. Khandelwal said the partnership could strengthen India’s position in the global digital economy.
How much capacity sits behind the project?
STTGDC first announced plans to develop in Noida in 2021. STT Noida 2 offers 22MW, Noida 1 offers 17MW, and Noida 3 is under development with 28MW planned.
The GPU deployment will use the STTGDC site as its infrastructure base, while Velocis supplies Neysa’s AI cloud platform. The arrangement links the 22MW facility to a platform intended for production training and inference workloads.
Who backs Neysa?
Neysa, founded in 2023, positions itself as a sovereign cloud player for India. The company has secured backing from:
- Blackstone
- Teachers’ Venture Growth
- TVS Capital
- 360 ONE
- Nexus
Separately, in April 2025, Neysa said it was teaming up with NTT Data and the Telangana government to establish a 400MW GPU cluster totaling 25,000 GPUs. The plan covers nearly three times the GPU count of Noida phase one.
What does STTGDC bring to the deal?
In February, investment firm KKR and Singaporean telecoms operator Singtel fully acquired STT GDC from ST Telemedia. The business has since adopted the STTGDC name. That transaction placed KKR and Singtel in control of the operator now hosting Neysa’s deployment.
STTGDC’s India division has more than 400MW of critical IT load in operation and development across 30 facilities in ten cities. The Noida project therefore sits within a broader portfolio spanning multiple operating and development sites.
Sanghi said Neysa is building for the long term to serve Indian businesses and global compute customers seeking to add India to their infrastructure footprint. He said the Noida project forms part of an approach that plans compute, platform and data-center infrastructure together.
Sanghi said: “Our customers need more than access to GPUs. They need capacity they can plan around, reliable operations, and room to scale as their workloads evolve.”
The planned deployment brings GPU infrastructure and Velocis together to support AI workloads in production. For cloud buyers, the immediate variables to track are the 9,000-GPU ceiling, the 22MW host site and the stated Q1 2027 go-live window.
via Data Center Dynamics (Source)
More from Elena Vasquez
Show full bio
Senior reporter covering people, launches and day-to-day practice in hardware layer of artificial intelligence at Die Signal.
498 articles
Same lot · LOT-C204
- DSG-582820nmMeta Leases AI Compute Capacity at Firmus' Southeast Asia Data Centers
- DSG-112814nmOracle Lines Up Nvidia, Google and Cerebras AI Chips for Its Cloud
- DSG-397210nmKarman software runs 16 Nvidia B300 servers on ten-server power budget
- DSG-459210nmOpenAI Hires Google Data Center Sourcing Lead Allyson Fealtman
- DSG-511928nmNvidia, Firmus plan 170,000-GPU data center in Indonesia