Lambda

What Lambda does

Lambda is a privately-held AI cloud infrastructure provider founded in 2012 in San Francisco (now headquartered in San Jose, California) that operates single-tenant, NVIDIA-aligned GPU clusters for AI training and inference at scale. The company's product stack is structured in three tiers: Instances (1-8 GPUs on pay-by-the-minute billing with no egress fees), 1-Click Clusters (16-2,000+ GPU NVIDIA HGX B200/H100 reservations deployable in minutes via a self-serve dashboard), and Superclusters (4,000-165,000+ GPUs on 3-5 year contracts with co-engineering and managed orchestration). Infrastructure spans 320+ MW leased across 20+ U.S. data centers via colocation partners (Aligned Data Centers, EdgeConneX, Prime Data Centers, ECL), with a 3 GW compute capacity target by 2030. Lambda holds SOC 2 Type II, ISO 27001/27017/27701/22301, CCPA, GDPR, and LGPD compliance and operates air-gapped Supercluster deployments for U.S. federal agencies (Air Force, Navy, DoE, Army, USSF, CENTCOM).

The platform is built on NVIDIA reference architectures (HGX B200, HGX B300, GB300 NVL72, VR200 NVL72, H100) with NVIDIA Quantum-2 InfiniBand and NVLink networking, direct-to-chip liquid cooling, integrated tiered storage, and an orchestration layer supporting managed Kubernetes, managed Slurm, dstack, and SkyPilot. Lambda ships its own curated ML software stack (Lambda Stack) pre-installed across all systems. The company maintains NVIDIA Exemplar Cloud status, is a 7-time NVIDIA Partner of the Year, and achieved MLPerf Training v6.0 fastest LLM training on GB300 NVL72 plus the first audited STAC-AI LANG6 result on HGX B200. Strategic customers include Microsoft (multibillion-dollar multi-year deal), Hudson River Trading, Orange, and government agencies; named startup and research customers include Kodiak, Pika, Iambic, Genesis, Meshy, and fal.

Lambda monetizes through usage-based instance pricing ($0.69-$6.99/GPU/hr), reservation-based 1-Click Cluster pricing ($5.54-$9.86/hr), and quote-based multi-year Supercluster contracts, with a managed-services layer bundled in. Distribution is hybrid: self-serve PLG for developers and startups, enterprise field sales for hyperscalers and frontier labs, and a CRADA-based channel for federal agencies. The capital base consists of a $1.5B+ Series E (Nov 2025) co-led by TWG Global and USIT, a $480M NVIDIA equity investment, a $1B upsized syndicated credit facility led by J.P. Morgan (May 2026), a $1.5B sale-leaseback arrangement, and prior equity rounds totaling ~$393M. Leadership includes former Sprint CEO Michel Combes as CEO (effective May 2026), co-founder Stephen Balaban as CTO, co-founder Michael Balaban as CPO, former AT&T Communications CEO John Donovan as Chairman, and ex-AWS/Sun/Snap COO Jerry Hunter as Vice Chairman. The company was reported in January 2026 to be in talks to raise $350M ahead of a planned IPO.

Name

Lambda

Legal name

Lambda, Inc.

Website

https://lambda.ai

Company type

Private

Founded year

2012

Operating status

Operating

Headcount range

501–1,000 employees

Short description

Lambda is a privately-held AI cloud infrastructure provider operating single-tenant NVIDIA GPU clusters — Instances, 1-Click Clusters, and Superclusters — for AI training and inference, serving hyperscalers (notably Microsoft), frontier AI labs, regulated enterprises, U.S. federal agencies, and AI startups.

Ownership category

akta.pro rank

Product category

GPU Cloud Infrastructure

NAICS

Computer Systems Design and Related Services (54151), Computing Infrastructure Providers, Data Processing, Web Hosting, and Related Services (518), Software Publishers (51321)

SIC

Services-Computer Integrated Systems Design (7373), Services-Prepackaged Software (7372), Services-Computer Programming, Data Processing, Etc. (7370)

akta.pro primary industry

AI Compute Cloud & GPU-as-a-Service (HDAAAAAK)

akta.pro secondary industries

GPU-Accelerated & AI Training/Inference Servers (HDACABAG), AI Server Systems & HGX/Accelerator Platforms (HDAAAAAB), AI Accelerators (GPUs/TPUs/NPUs/ASICs) (HDAAAAAA), AI Compiler, Runtime & Kernel Optimization Software (CUDA/ROCm/XLA, graph compilers) (HDAAAAAI)

Keywords

• GPU cloud infrastructure
• AI supercomputing services
• Neocloud AI compute
• AI training infrastructure
• NVIDIA GPU clusters

Headquarters

HQ city

San Jose

HQ country

United States

HQ region

North America

Offices9 records

Markets served

GTM type

B2B

Offering type

Software

Cost components

Technology or R&D, Infrastructure, Supply Chain, Personnel, Marketing or Sales, Operations

Revenue model

1. On-demand GPU Instances: Pay-by-the-minute self-serve GPU instances ranging from $0.69/hr (Quadro RTX 6000) to $6.99/GPU/hr (NVIDIA B200 SXM6) for 1 to 8 GPU configurations, no ingress/egress fees, available in minutes.
2. 1-Click Cluster reservations: Multi-week to multi-year reservations for 16-2,000+ GPU NVIDIA HGX B200 or H100 clusters; on-demand 2-week-to-1-year pricing ranges from $5.54/hr (H100 256-GPU) to $9.86/hr (B200 16-GPU), with multi-year contracts via sales.
3. Supercluster multi-year contracts: Long-term (3-5 year) single-tenant Supercluster contracts for 4,000-165,000+ NVIDIA GPUs at hyperscale, with co-engineering, managed orchestration, and dedicated infrastructure. Quote-based pricing.
4. Managed services (Kubernetes, Slurm, Cloud): Managed cluster operations including control plane deployment, updates, monitoring, and SLA-backed job scheduling bundled with Supercluster and 1-Click Cluster deployments.
5. Hardware sales (legacy): Hardware and ancillary services sales under separate Hardware Terms of Sale; legacy revenue stream as the company has shifted to a cloud/infrastructure-as-a-service primary model.

Pricing tiers

Go-to-market motion6 records

Distribution channels7 records

Marketing channels10 records

Core offering

Lambda builds and operates single-tenant AI cloud infrastructure (Superclusters, 1-Click Clusters, and Instances) running on NVIDIA GPU architectures including HGX B200, GB300 NVL72, and VR200 NVL72. The company sells dedicated GPU compute capacity for AI training and inference workloads, with managed orchestration (Kubernetes/Slurm), pre-installed Lambda Stack ML software, and direct-to-chip liquid-cooled AI Factories deployed across U.S. data centers.

Differentiator

Problem solved

Functional benefit

Brands

• Lambda Stack: Curated deep learning software package pre-installed on Lambda products, bundling PyTorch, TensorFlow, Keras, JAX, CUDA, and related tools.
• Lambda Cloud

Products and services

• Superclusters Single-tenant, shared-nothing AI cloud providing clusters from 4,000 to 165,000+ NVIDIA GPUs on 3+ year contracts. Purpose-built for large-scale AI training and inference with NVIDIA-aligned reference architectures, full observability, and managed orchestration.
• 1-Click Clusters Self-serve, production-ready NVIDIA HGX B200 or H100 clusters from 16 to 2,000+ GPUs, deployable in minutes via dashboard for short-term or multi-year reservations. Fully optimized for distributed AI training, fine-tuning, and inference workloads.
• Instances On-demand 1 to 8 GPU instances across NVIDIA B200 SXM6, H100 SXM, A100 SXM/PCIe, GH200, A6000, A10, V100, and RTX 6000. Pricing from $0.69/hr to $6.99/GPU/hr with pay-by-the-minute billing, zero egress fees, and instant access for training, fine-tuning, and serving models.
• AI Factories (Contiguous and Distributed) Modular AI factory infrastructure including Contiguous campuses (75MW+ for large clusters and mission-critical workloads) and Distributed sites (5-75MW in strategic metros for sovereign-AI compliance and low-latency). Combines high-density power, direct-to-chip liquid cooling, and N+1 to N+N redundancy.
• Lambda Bare Metal Instances Full hardware control with API-driven operations for rack-scale NVIDIA GPU, CPU, scale-up networking fabric, and liquid cooling integration at the unit-of-AI-compute level.

Quantifiable outcome

• MLPerf Training v6.0: 18.7% improvement on NVIDIA GB300 NVL72 Llama 3.1 8B submission vs prior Lambda result — fastest LLM training.
• +6 more outcomes

Named customers17 records

Segments6 records

Ideal customer profiles5 records

Technology focussed Yes

API detail

Has API

Yes

API docs

API detail

Core technology

AI maturity

App detail

Integration8 records

AI capability12 records

Feature8 records

Partnerships

18 partnerships are on record, tiered flagship launch partner, major customer-partner, strategic open-source partner, major infrastructure partner, core ecosystem partner, core hardware partner, strategic infrastructure partner, flagship strategic customer and partner, flagship strategic partner, sustainability showcase partner and core mlops partner.

• NVIDIA (CPO switch early access)flagship launch partnerStrategic or Co-development Partner · 3 June 2026Lambda received early access to NVIDIA's upcoming Quantum-X Q3450-LD co-packaged optics (CPO) networking switch, now in full production for use in Lambda's GB300 NVL72 systems. Switch features 144x 800G ports delivering 115.2 Tbps with substantial power savings vs traditional pluggable transceivers.
• Hudson River Trading (HRT)major customer-partnerStrategic or Co-development Partner · 20 May 2026Lambda announced a partnership with Hudson River Trading (HRT) on May 20, 2026 to provide NVIDIA accelerated computing infrastructure including HGX B200 systems for HRT's quantitative trading research and development. Lambda delivers full-stack architecture enabling HRT researchers to train models and simulate trading strategies at scale.
• Red Hatstrategic open-source partnerTechnology or Integration · 24 March 2026Lambda is a partner in llm-d, a distributed AI inference framework jointly developed by Red Hat, Google Cloud, IBM Research, CoreWeave, and Nvidia. llm-d joined the CNCF as a Sandbox project in March 2026, with Lambda alongside AMD, Cisco, Hugging Face, Intel, and Mistral AI as subsequent partners.
• Google Cloudstrategic open-source partnerTechnology or Integration · 24 March 2026Lambda is a partner in llm-d, a CNCF Sandbox project jointly developed with Google Cloud, Red Hat, IBM Research, CoreWeave, and Nvidia. Lambda's multi-cloud platform also supports Google Cloud deployments.
• IBM Researchstrategic open-source partnerTechnology or Integration · 24 March 2026IBM Research is a co-launch partner of llm-d, a distributed AI inference framework now under CNCF with Lambda as a partner. Lambda's multi-cloud platform also supports AWS, Azure, GCP, and OCI deployments.
• CoreWeavestrategic open-source partnerTechnology or Integration · 24 March 2026CoreWeave is a co-launch partner of llm-d, a distributed AI inference framework now under CNCF with Lambda as a subsequent partner. Both companies operate neocloud AI infrastructure businesses.
• Hugging Facestrategic open-source partnerTechnology or Integration · 24 March 2026Hugging Face is a partner in llm-d, a CNCF Sandbox distributed AI inference framework alongside Lambda. Lambda's inference model catalog includes a wide range of HuggingFace-hosted open-source models.
• Aligned Data Centersmajor infrastructure partnerChannel Partner/ Reseller/ Distributor · 11 March 2026Aligned Data Centers is constructing a 425,000-square-foot, $700 million data center facility in Plano, Texas for Lambda. Aligned's newest Dallas-Fort Worth facility DFW-04 is a liquid-cooled AI & cloud data center occupied by Lambda for next-generation AI infrastructure (announced May 7, 2025).
• Open Compute Project (OCP)core ecosystem partnerStrategic or Co-development Partner · 27 January 2026Lambda joined the Open Compute Project (OCP) Advisory Board on January 27, 2026. This reflects Lambda's belief that the future of AI infrastructure depends on proven designs becoming open, repeatable, and scalable across the industry.
• Supermicrocore hardware partnerOEM/ Whitelabel/ Licensing Partner · 1 January 2026Lambda AI factories are engineered in partnership with Supermicro alongside NVIDIA and Dell Technologies. Supermicro builds the rack-scale systems (e.g., 142 kW NVIDIA GB300 NVL72 systems) used in Lambda's AI factories.
• Dell Technologiescore hardware partnerOEM/ Whitelabel/ Licensing Partner · 1 January 2026Lambda AI factories are engineered in partnership with Dell Technologies alongside NVIDIA and Supermicro, providing enterprise-grade server infrastructure for Supercluster and 1-Click Cluster deployments.
• Equinixstrategic infrastructure partnerChannel Partner/ Reseller/ Distributor · 1 January 2026Lambda and Equinix mentioned as partners in building and expanding AI factories in Lambda's AI infrastructure page news section.
• Prime Data Centersmajor infrastructure partnerChannel Partner/ Reseller/ Distributor · 13 November 2025Prime Data Centers and Lambda partnered to power the next era of superintelligence with AI-optimized infrastructure in Southern California. New deployment at LAX01 in Vernon (Vernon's first AI-ready data center) delivers purpose-built, NVIDIA Blackwell infrastructure (announced November 13, 2025).
• Microsoftflagship strategic customer and partnerStrategic or Co-development Partner · 3 November 2025Lambda announced a multibillion-dollar multi-year agreement with Microsoft on November 3, 2025 to deploy tens of thousands of NVIDIA GPUs as mission-critical AI cloud compute at scale. Lambda is also 'Microsoft-backed' per multiple press references.
• NVIDIA (Exemplar Cloud)flagship strategic partnerTechnology or Integration · 9 October 2025NVIDIA Exemplar Cloud partner — Lambda achieved NVIDIA Exemplar Cloud status as one of the first cloud providers validated for mission-critical AI training workloads at scale. Lambda is a seven-time NVIDIA Partner of the Year and a launch partner for the latest generations of NVIDIA GPUs (VR200 NVL72, GB300 NVL72, HGX B300).
• ECLsustainability showcase partnerStrategic or Co-development Partner · 23 September 2025Lambda and ECL brought the first hydrogen-powered NVIDIA GB300 NVL72 systems online in September 2025. Supermicro-built 142 kW systems meet zero-emissions energy at ECL's Mountain View facility, where Lambda has doubled its footprint.
• EdgeConneXmajor infrastructure partnerChannel Partner/ Reseller/ Distributor · 21 August 2025EdgeConneX and Lambda announced on August 21, 2025 the build of 30+ megawatts of AI factory data center infrastructure in Chicago and Atlanta. EdgeConneX is a data center solutions company providing AI infrastructure for hyperscale, neocloud, and enterprise workloads.
• Weights & Biasescore mlops partnerTechnology or Integration · 1 May 2023Weights & Biases formed a strategic partnership with Lambda in May 2023 to provide GPU cloud infrastructure and MLOps tools for AI and deep learning workloads. The partnership integrates Lambda's GPU compute with W&B experiment tracking and model management.

Scale indicators14 records

Recent moves14 records

Expansion highlights7 records

Direct peers

• CoreWeave: Closest direct neocloud peer: NVIDIA-backed ($2B equity) GPU cloud built on similar NVIDIA HGX/GB/VR200 architectures, offering on-demand instances, reserved clusters, and large multi-year Supercluster deals to hyperscalers and AI labs.
• Crusoe: Neocloud competitor building large-scale AI factories with NVIDIA GPUs and energy-flared/stranded power, pursuing the same frontier AI lab and hyperscaler customers with multi-gigawatt ambitions.
• Nebius: AI-centric neocloud spun out of Yandex with $2B+ NVIDIA backing, operating NVIDIA GPU clusters across Europe and the U.S. and directly competing for the same frontier model training workloads.

Broad incumbents

• Amazon Web Services (AWS): Hyperscaler with EC2 P5/P5e GPU instances, Bedrock, and Trainium silicon; competes with Lambda for the same enterprise and frontier AI lab workloads but as part of a much broader cloud portfolio.
• Google Cloud Platform: Hyperscaler offering A3 GPU instances (H100), TPU, and Vertex AI; Lambda supports GCP deployments in its multi-cloud platform while GCP competes for the same AI training and inference workloads.
• Microsoft Azure: Hyperscaler that is simultaneously Lambda's largest disclosed customer (multibillion-dollar GPU deal) and a competitor via Azure ND H100/B200 instances and Azure OpenAI Service — central to Lambda's upside and concentration risk.

Emerging players

• Together AI: AI cloud focused on open-source and inference workloads, running large NVIDIA GPU clusters (some in partnership with Crusoe); overlaps with Lambda on LLM serving and developer-led customers, though more software/platform oriented.
• Paperspace (DigitalOcean): DigitalOcean-owned GPU cloud offering on-demand H100/A100 instances and notebooks for AI developers; overlaps with Lambda's self-serve Instances product and startup/research customer segment.
• RunPod: GPU cloud platform offering on-demand and serverless GPU instances for AI training and inference; targets the same self-serve developer and startup segment as Lambda Instances with comparable transparent pricing.
• Vultr: Independent cloud provider offering GPU instances (H100, L40S, etc.) on a global footprint; competes with Lambda's Instances product for self-serve AI developers and smaller training workloads.

Market position

Strengths5 records

Weaknesses5 records

Competitive moat6 records

Key risks6 records

Key highlights7 records

Customer concentration

• Xx.com/LambdaAPI
• GitHubgithub.com/LambdaLabsML
• YouTubeyoutube.com/@lambda-ai
• LinkedInlinkedin.com/company/lambda-cloud
• Facebookfacebook.com/lambdalabs

Compliance8 records

Revenue estimate

Valuation estimate

Number of profiles

Profiles12 records

Funding overview

Funding rounds17 records

Investors52 records

Funding detail is available on the Subscription and Enterprise plan.Contact sales →

M&A

Investments5 records

M&A and investment is available on the Subscription and Enterprise plan.Contact sales →

Frequently asked questions about Lambda