Sustainably enabling AI, everywhere.
Enabling frictionless access to powerful AI compute and denser, more sustainable data centres through the world’s first scaled immersion cooled AI computing platform.
Firmus has pioneered scaled immersion cooling, to solve the challenge of accessing the benefits of this breakthrough technology, today.
Emitting up to 50% less CO2 when in use, with a corresponding decrease in power consumption and costs, Firmus’ Cube platform is built to host hyperscale-level amounts of high-performance compute in the data centre, at the edge, or for the cloud. Through our partnership with ST Telemedia Global Data Centres (STT GDC), Firmus and STT GDC are rolling out the world’s first fleet of Sustainable AI Factories, starting in Singapore and India.
As an upgrade to air cooled facilities or stand-alone edge, the Cube computing platform enables the frictionless adoption of immersion cooling to make legacy DC’s more sustainable and capable, plus deploy AI compute to the edge as robust and reliable modular Sustainable AI Factories.
- [FRICTIONLESS USE]
FULL STACKAs a full platform of infrastructure & compute, Cube enables access to immersion's breakthrough sustainability & economics without detailed knowledge transfer or operational interruption. Firmus’ tech stack extends to the compute and network fabric to plug and play cutting edge immersion technology today.
- [RELIABLE & ROBUST]
N+1A modularised, full N+1 platform, aligning immersion technology with industry standards for reliability and robustness. Within the DC, Cube adapts to existing water & chiller loops and power distribution. At the edge, Cube is deployed with all backup and cooling required for N+1 operation.
- [POWERFUL HARDWARE]
AI & MLHosting up to 130kW per 42RU, Firmus' Cube platform is built to perform, able to host the most powerful AI servers and GPUs with ease. Tested and designed with industry leaders including NVIDIA, Firmus’ metal options are customised across a full spectrum of compute uses.
- [MORE CAPABLE]
DC UPGRADEUpgrades legacy DCs to reduce overall PUE, increase density and ROI, and unlock the ability to host high powered AI metal servers. Firmus’ Cube platform is a frictionless approach to making data centres more sustainable and capable without scrapping existing infrastructure or investment.
- [BREAKTHROUGH SUSTAINABILITY]
CO2Sub 1.05 PUE in any environment cuts CO2 emissions in all use cases by as much as 50% against the global standard for data centre efficiency. Cube enables more sustainable computing in any location.
AI CLOUDThrough Sustainable Metal Cloud, Firmus demonstrates the power of Cube’s full-stack metal orchestration and AI platform, built to NVIDIA's exacting architectures for deep learning & Omniverse workloads.
- [AI CLOUD]
Cube enables Sustainable Metal Cloud - the world's sustainable, lower cost AI cloud.Visit Sustainable Metal Cloud
The power of Cube’s scaled immersion computing platform is paired with NVIDIA deep learning A100 & H100 GPUs and fast network fabric to deliver a sustainable, lower cost AI computational cloud like no other. Sustainable Metal Cloud (SMC) is bringing a hyperscale cluster of deep learning and Omniverse compute to Asia, India & Europe, delivered from ST Telemedia Global Data Centres’ (STT GDC) Cube-enabled sites.
SMC democratizes access to AI like nothing else before it, unlocking the possibility of this new tool in a sustainable, cost-effective and powerful way.
- [DATA CENTRE]
Add capability & sustainability to any DC, end-to-end with AI compute
ST Telemedia Global Data Centres (STT GDC) and Firmus are partnering to build the world’s first fleet of Sustainable AI Factories. STT GDC’s global footprint with over 1GW of IT load are being upgraded to prepare for the era of AI and the Omniverse. Starting with Singapore and India, and scaling into other regions, Cube infrastructure is being deployed with deep learning and Omniverse compute, allowing access to these resources in typically constrained regions.
Cube-enabled DCs allow customers to access wholesale, powerful AI compute to open new regions, or consume hyperscale & sustainable high performance metal for their own stacks. Each Cube-enabled DC markedly lowers PUE and TCO of operating high-performance metal like nothing else. Make an enquiry to learn more.
Delivers AI to the edge for critical zero-latency deployments
Cube 300 delivers 300KW of hosting capacity over 145RU. This modular Sustainable AI Factory delivers the breakthrough engineering that is common to all Cube infrastructure: <1.05 stand-alone PUE, a 12 x smaller land footprint, a TUE of less that 1.15, and N+1 reliability.
With the same GPU capacity as 1.5 NVIDIA DGX SuperPODs, in a sustainable, reliable, go-anywhere platform – our modular solution delivers AI to the edge quickly and with high performance.
Get in touch
Provide more information on what you’re after, and we’ll be in contact within the next 24 hours.