Skip to main content

Think launches Think Grid™, a new alternative to hyperscaler AI compute

Source Decision Insights Signals Published August 31, 2026 Mentions 2 companies

Companies mentioned

Think said it launched Think Grid, a hosted AI compute service that provides dedicated capacity from Think Fabric without customers funding dedicated infrastructure upfront. The company said the service began in Riyadh and then expanded globally.

Think Grid was hosted in a Tier III facility in Riyadh and delivered as a monthly subscription for dedicated capacity. Think said customers could either deploy Think Fabric within their own infrastructure or consume it as a fully managed, hosted service, with Think Grid handling hardware and infrastructure operations.

At launch, Think Grid SuperNode combined four NVIDIA PRO 6000 Blackwell accelerators with 384GB of total VRAM, along with Think’s ILM orchestration software. Think said ILM pooled GPU memory into a unified memory space and orchestrated workloads across available compute so multiple models could share the infrastructure rather than leaving individual accelerators underutilised.

Think said ILM-managed nodes ran multiple production models in production testing on hardware that conventional deployments dedicate to one. The company said each all-inclusive monthly rate covered a dedicated bare-metal node and did not include separate storage, support, or data egress charges. Ahmed AlSharif, Founder and CEO of Think, said, “For too long, organisations have effectively had two choices when they need serious AI compute: make a significant capital investment in infrastructure of their own, or rent capacity from a hyperscaler and accept the complexity and variable costs that come with it,” said Ahmed AlSharif, Founder and CEO of Think. “Think Grid creates another option. By providing the Think Fabric as a service, we’re giving customers dedicated, state-of-the-art AI compute, with the software, hardware and operations bonded together under one predictable monthly rate.”

Think said pricing comparisons were based on published August 2026 hyperscaler pricing and that Think Grid’s all-inclusive monthly rate was up to 27% lower than comparable dedicated Blackwell AI compute configurations, before accounting for storage, data egress, and support charges that were typically billed separately.

Press release, provided by Cision on behalf of ThinkParQ. Read the original.

Graph Connections

2 companies named across 5 categories, one of 566 sources referencing Nvidia. Previous coverage: Quantum and enterprise AI deployments, SIEM and network automation updates - Week of August 31, 2026 (August).