Skip to main content

Positron

Category: AI Infrastructure

A semiconductor company developing specialized, energy-efficient hardware and software architectures specifically optimized for Transformer-based AI inference. Positron was founded in 2023. The company is led by Mitesh Agrawal. Based in Reno, USA. Team size: 20-50. Total funding raised: $250.0M. Latest round: Series B. Key investors include ["Arm Holdings","Qatar Investment Authority (QIA)","Valor Equity Partners","Atreides Management","DFJ Growth"].

Founded
2023
Headquarters
Reno, USA
Team size
20-50
Total funding
$250.0M

Value proposition

Delivers significantly higher throughput and lower energy consumption for LLM inference compared to general-purpose GPUs, effectively breaking the 'NVIDIA bottleneck' for scaling AI models.

Products and solutions

["Transformer-optimized Inference Accelerators","Energy-efficient AI Silicon Architecture","Hardware-Software Co-design Stack","High-performance AI Compute Clusters"]

Unique value

Unlike general-purpose GPUs, Positron's architecture is 'Transformer-native,' designed from the ground up to handle the specific mathematical and memory-bandwidth requirements of Large Language Models.

Target customer

Hyperscale cloud providers, AI infrastructure companies (e.g., Lambda), enterprise data centers, and LLM developers.

Industries served

["Semiconductors","Artificial Intelligence","Cloud Computing","Data Centers"]

Technology advantage

Combines high-level functional programming principles (via Chief Scientist Edward Kmett) with energy-efficient hardware design to optimize data flow. Business advantage includes a leadership team with deep NVIDIA/Lambda scaling experience and strategic alignment with the Arm ecosystem.

How they differentiate

Positron utilizes a 'Transformer-native' architecture designed from the ground up for LLM inference, rather than general-purpose compute. By integrating high-level functional programming principles into hardware-software co-design, it achieves significantly higher throughput and energy efficiency for transformer models compared to standard GPUs.

Main competitors

["NVIDIA","Groq","Etched","Cerebras Systems"]

Key partnerships

["Arm Holdings (Strategic Investor and Architecture Partner)","Qatar Investment Authority (Financial/Strategic Partner)","Lambda (Strategic relationship via executive leadership)","Valor Equity Partners"]

Notable customers

["Lambda (Strategic Partner)","Hyperscale Cloud Providers","Enterprise Data Centers"]

Major milestones

["Company founded in 2023 by Thomas Sohmers and Edward Kmett","Appointed Mitesh Agrawal (former NVIDIA VP and Lambda COO) as CEO in January 2025","Closed $230M Series B funding round led by Arm and QIA in February 2026"]

Growth metrics

Scaled from stealth to a $230M Series B within three years; leadership team includes former executives who scaled Lambda to $500M ARR.

Market positioning

High-performance AI silicon challenger targeting the 'NVIDIA bottleneck' in enterprise and hyperscale data centers.

Geographic focus

North America, Middle East (via strategic QIA backing), and Europe (via Arm partnership).

Patents and IP

Proprietary architecture for transformer acceleration and energy-efficient data movement (specific public filings not disclosed in brief).

About Mitesh Agrawal

Mitesh Agrawal is a seasoned technology executive with over 20 years of experience in semiconductors and AI scaling. Before becoming CEO of Positron, he served as the Chief Operating Officer at Lambda, where he was instrumental in scaling the AI cloud provider's revenue from $500K to $500M ARR and securing over $1B in financing. His foundational experience includes a 4-year tenure at NVIDIA as VP/GM of Cloud and AI Software and 7 years at Qualcomm as a Staff Engineer. He holds an MBA from UC Berkeley Haas and an MS in Electrical and Computer Engineering from UT Austin.

Official website: