← All companies

Neoclouds / Company profile

TensorWave

What TensorWave does

TensorWave is an AMD-exclusive AI cloud provider (a “neocloud”) that sells high-performance compute and cluster infrastructure for AI training, fine-tuning, and inference. The company positions itself around a fully AMD GPU and ROCm-based stack to provide an alternative to NVIDIA-centric supply, with infrastructure designed to scale from distributed training to high-throughput production inference.

TensorWave’s platform combines GPU hardware (AMD Instinct accelerators), storage and networking, and managed orchestration. It emphasizes (1) memory- and network-intensive AI workloads, (2) Kubernetes at scale for production inference/training, and (3) HPC-style scheduling via a Slurm-oriented approach for large-scale training and fine-tuning. On the inference side, the company highlights managed Kubernetes, RoCEv2-based clustering interconnects, and compliance claims (SOC 2 Type II, ISO 27001, and HIPAA) tied to its inference offering.

From a go-to-market standpoint, TensorWave appears to target AI labs, enterprises, and cloud/AI builders who want managed access to AMD accelerators and production-oriented cluster operations (monitoring, security, fleet management, and expert support).

News

Company record

Aug 11, 2026 · Official · TensorWave blog (reposting Fermi announcement)Fermi executes its first binding lease at Project Matador with TensorWave (222 MW; expansion rights to >650 MW)

Fermi announced a binding customer lease with TensorWave via TensorWave TEX1, LLC for a facility supported by 222 MW of total facility power, with an initial 15-year term and approximately $6.5 billion in total contracted revenue over the initial term for phase one, plus expansion rights to more than 650 MW across two additional data centers.

Jul 23, 2026 · Official · TensorWave blogTensorWave adopts AMD Helios rackscale solution to advance frontier-scale inference and training

TensorWave announced it adopted AMD’s Helios rackscale solution powered by AMD Instinct MI455X GPUs, describing it as a unified rack-scale building block combining compute and networking for high-volume inference and frontier-model training/fine-tuning.

Jun 18, 2026 · Official · TensorWave blogTensorWave publishes a power-and-infrastructure perspective on modern AI data centers (megawatts as the limiting constraint)

TensorWave published an article arguing that electricity sourcing and delivery (megawatts, cooling, networking, and compute integration) increasingly drive AI infrastructure capacity and performance outcomes.

Jun 10, 2026 · Reporting · DataCenterDynamicsTensorWave closes $350M Series B at $1.55B valuation to expand AMD-powered AI infrastructure

TensorWave secured a $350 million Series B co-led by Magnetar and AMD Ventures, valued at $1.55 billion, including further AMD Instinct MI355X cluster deployments.

Jun 03, 2026 · Official · TensorWave blogTensorWave discusses agent-driven development for AI infrastructure operations

TensorWave published a post summarizing a Beyond Summit session on how AI systems are changing debugging/profiling/testing/optimization workflows and how infrastructure tooling may evolve to support agentic interaction patterns.

May 22, 2026 · Official · TensorWave blogTensorWave: Spectral Compute runs unmodified CUDA workloads on TensorWave infrastructure (up to 25.7× faster performance)

TensorWave published a collaboration post describing a benchmarking/validation workflow run on TensorWave’s AMD Instinct MI355X infrastructure to support unmodified CUDA applications via Spectral Compute’s SCALE toolkit, including reported performance and managed-memory migration improvements.

May 19, 2026 · Official · TensorWave blogWafer reaches #1 public inference performance for Qwen3.5-397B on AMD (MI355X) using TensorWave infrastructure

TensorWave published a post referencing Wafer’s inference performance claims and describing how TensorWave and AMD-native inference tooling (ATOM) were involved in the described achievement.

Apr 06, 2026 · Official · TensorWave blogTensorWave + Modular: deploy AMD inference beyond NVIDIA (MAX inference framework)

TensorWave published a partnership post with Modular describing deployment of Modular’s MAX inference framework on TensorWave’s AMD MI355X infrastructure, with reported production and cost-per-output positioning.

Show 4 earlier updates
Mar 02, 2026 · Official · TensorWave blogTensorWave and AT&T (with AMD) to build a foundation for telco-grade models

TensorWave published a post describing a partnership with AT&T and AMD aimed at enabling deployment foundations for telco-grade AI models in regulated/operational environments.

Feb 25, 2026 · Reporting · Credo/Business Wire PDF hosted by Credo IRCredo and TensorWave partner to deploy ZeroFlap active electrical cables and optics across next-generation AI clusters

Credo announced a collaboration with TensorWave to deploy Credo’s ZeroFlap (ZeroFlap AECs and optics) across TensorWave’s next-generation AMD-based AI cluster infrastructure, positioning it around time-to-first-token and reliability goals.

Jan 14, 2026 · Official · TensorWave blogTensorWave and TECfusions expand AMD capacity across Tucson and Pennsylvania

TensorWave published a post describing capacity expansion tied to TECfusions deployments and referencing speed-to-capacity and additional MW builds on earlier phases.

Oct 08, 2024 · Official · TensorWave blogTensorWave raises $43M SAFE funding (largest in Nevada startup history)

TensorWave announced the closing of a $43 million SAFE round led by Nexus VP, with participation from Maverick Capital, Translink Capital, Javelin Venture Partners, StartupNV, and AMD Ventures.

Source map · 2 recurring channels · 11 references

Still resolving: Newsroom · Official X

What it builds

TensorWave AI Cloud (AMD Instinct-based infrastructure)

An AMD-exclusive AI cloud for training, fine-tuning, and inference that is positioned around AMD Instinct GPUs and ROCm-based acceleration for production-scale workloads.

source ↗
Bare Metal

Bare-metal access to AMD Instinct accelerators as part of the company’s infrastructure offering for AI workloads.

source ↗
Managed Kubernetes

Managed Kubernetes designed for deploying production GPU workloads across large fleets, positioned for AI training, fine-tuning, and inference.

source ↗
Managed Slurm

A scheduling layer positioned for large-scale training and fine-tuning workloads using Slurm concepts integrated with a Kubernetes-native orchestration approach.

source ↗
AI Model Training (training infrastructure)

Training infrastructure including fast, predictable distributed-training communication (RoCEv2 fabric / ultra Ethernet-ready design), checkpointing support, and managed scheduling for large-scale training.

source ↗
AI Inference

High-performance inference offering positioned for speed and scale, including Kubernetes at scale, cluster-scale inference across RoCEv2-based interconnects, and claims of compliance coverage tied to the service.

source ↗
AI Fine-Tuning

A fine-tuning offering for custom model training on TensorWave’s AMD GPU infrastructure.

source ↗
TensorWave Enterprise Suite

A full-stack production offering described as an enterprise platform for deploying AI workloads on TensorWave infrastructure.

source ↗

Leadership

Milestones & partnerships

Aug 11, 2026
Fermi announces binding lease agreement with TensorWave for 222 MW (phase one) with expansion rights

TensorWave reposts the Fermi announcement detailing phase-one 222 MW delivery in 2027 timeframe and expansion rights to >650 MW.

Fermi America (Fermi Inc.)
Jul 23, 2026
Adopted AMD Helios rackscale solution

TensorWave announced adoption of AMD Helios rackscale solution powered by MI455X GPUs to support frontier-scale inference and training workloads.

AMD
Jun 10, 2026
Series B closes (co-led by Magnetar and AMD Ventures)

DCD reports TensorWave closed its $350M Series B at a $1.55B valuation, co-led by Magnetar and AMD Ventures, supporting MI355X cluster expansion.

Magnetar · AMD Ventures · Maverick Silicon · Nexus Venture Partners · Western Frontier
Apr 06, 2026
Partnership with Modular to run MAX on TensorWave AMD inference infrastructure

TensorWave published a post describing running Modular’s MAX inference framework on TensorWave’s AMD MI355X infrastructure.

Modular
Feb 25, 2026
Partnership with Credo for ZeroFlap active electrical cables and optics

Credo announced deploying its ZeroFlap (AECs and optics) across TensorWave’s next-generation AI cluster infrastructure.

Credo Technology Group Holding Ltd
Jan 14, 2026
Capacity expansion with TECfusions (Tucson & PA)

TensorWave describes expanding AMD capacity builds with TECfusions and references MW progress tied to earlier phases.

Tecfusions
Jul 15, 2025
Launched an AI training cluster with 8,192 AMD Instinct MI325X GPUs online (reported by DCD)

DataCenterDynamics reports TensorWave launched an AI training cluster online with 8,192 AMD Instinct MI325X GPUs.

Oct 08, 2024
Raised $43M SAFE to expand capacity and launch new inference platform (company-described)

TensorWave announced the closing of a $43M SAFE round led by Nexus VP to scale the team, secure data-center capacity, and deploy AMD Instinct MI300X GPUs; the post also references an inference-platform launch and MI325X readiness.

Nexus Venture Partners · Maverick Capital · Translink Capital · Javelin Venture Partners · StartupNV · AMD Ventures
Nov 01, 2023
Company established in Nevada (TensorWave Inc. founded in November 2023)

A Nevada economic development board packet states TensorWave Inc. was founded in November 2023.

Nevada Governor’s Office of Economic Development (GOED)