Trio AI

What Trio AI does

Trio AI Limited is a Hong Kong-headquartered AI infrastructure company that provides GPU-as-a-Service (GaaS) for training, fine-tuning, serving, and inferring AI models. The company was formally established as Trio AI Limited in Hong Kong SAR and operates from 46th Floor, Lee Garden One in Causeway Bay, with an additional AI Application R&D Center at Cyberport Hong Kong established in November 2024. Its target customers span HKSAR Government, enterprise verticals (finance, retail, life science, smart manufacturing), universities and research institutions, and AI-native startups across Hong Kong, the Greater Bay Area, and Southeast Asia.

The core technology is a proprietary heterogeneous GPU computing platform with a 1,000-GPU cluster launched on August 23, 2024, followed by Hong Kong's first Chinese high-performance 1,024-GPU cluster built with METAX using 400G InfiniBand, all-flash storage, and high-performance GPU servers optimized for large-model workloads. Layered on top is AIDC Platform Management Software for multi-brand GPU orchestration, a Model-as-a-Service offering providing access to pre-trained models, and Data Pools for cross-border data collection and storage. A stated point-to-point compliant cross-border platform links Hong Kong and Shanghai, with plans to extend into Southeast Asia, and the GPU footprint is positioned to scale beyond 10,000 GPUs.

Trio AI monetizes through three primary GaaS service models: Customized Deployment (fixed multi-year committed contracts for organizations with sustained GPU demand), On-Demand (pay-as-you-go usage-based billing for fluctuating workloads), and a GPU-for-Startups program that issues GPU compute in exchange for equity stakes. Supporting revenue streams include training services, inference services, and value-added data services across Hong Kong and Shanghai. Go-to-market is enterprise field sales supplemented by a strategic channel partnership with HKT Limited (MOU signed December 11, 2024) for reselling the Artificial Intelligence Super Computing (AISC) solution in Hong Kong, GBA, and Southeast Asia via HKT's 800 Gbps AI Superhighway, alongside event-driven demand generation through forums such as The Belt and Road Software Pavilion 2024. The company is privately held, discloses no funding rounds, and reports an employee headcount in the 1–10 range, with leadership including CEO Dr. David Chung, COO Ms. Du Feilu, and Chief Strategy Officer Maureen Ng.

Name

Trio AI

Legal name

Trio AI Limited

Website

https://aitrio.ai

Company type

Private

Operating status

Operating

Headcount range

1–10 employees

Short description

Trio AI is a Hong Kong-based AI infrastructure provider offering GPU-as-a-Service on heterogeneous GPU clusters to enterprises, government, universities, and AI startups across the Greater Bay Area and Southeast Asia.

Ownership category

akta.pro rank

Product category

AI Infrastructure / GPU Cloud Computing

NAICS

Computing Infrastructure Providers, Data Processing, Web Hosting, and Related Services (5182), Computer Systems Design and Related Services (54151)

SIC

Services-Computer Rental & Leasing (7377), Services-Computer Integrated Systems Design (7373)

akta.pro primary industry

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

akta.pro secondary industries

AI Server Systems & HGX/Accelerator Platforms (HDAAAAAB), AI Compute Virtualization & Scheduling (GPU virtualization, cluster schedulers) (HDAAAAAG), AI Networking for Clusters (InfiniBand/Ethernet/RDMA) (HDAAAAAE)

Keywords

• GPU cloud services
• AI infrastructure
• GPU-as-a-Service
• High-performance computing
• AI model training

Headquarters

HQ city

Causeway Bay

HQ country

Hong Kong SAR China

HQ region

Asia

Offices4 records

Markets served

GTM type

B2B

Offering type

Software

Cost components

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

Revenue model

1. GPU-as-a-Service - Customized Deployment: For organizations with substantial and consistent GPU needs, offering customized deployment with fixed committed contracts at guaranteed scale. Provides predictable cost, economies of scale, dedicated support, and resource availability assurance.
2. GPU-as-a-Service - On Demand: Pay-as-you-go consumption-based billing for fluctuating or unpredictable GPU requirements. Maximum flexibility allowing scaling GPU resources up or down based on project demands.
3. GPU for Startups Program: Exclusive program where startups access GPU resources in exchange for equity stake. Provides computing power without upfront cost while enabling portfolio companies to focus on product development.
4. Value-added Data Services: Data collection, storage and application services across Hong Kong, Shanghai and additional locations for cross-border clients and projects
5. Training Service: AI model training, fine-tuning and serving services with availability and economies that empower scale
6. Inference Service: Serving better inference and scale across GPUs as demand changes

Pricing tiers

Go-to-market motion2 records

Distribution channels3 records

Marketing channels5 records

Core offering

Trio AI provides GPU-as-a-Service (GaaS) in Asia, delivering proprietary access to heterogeneous GPU computing power through customized deployment contracts, on-demand consumption-based access, and a startup equity-exchange program. The company operates Hong Kong's first Chinese high-performance 1,024-GPU cluster and offers supporting services including model training, inference, Model-as-a-Service, and cross-border data services connecting Hong Kong and Shanghai.

Product overview

Trio AI is an AI infrastructure service provider offering GPU-as-a-Service (GaaS) in Asia. The core offering is heterogeneous GPU computing power accessible through multiple service models: customized deployment (fixed committed contracts), on-demand (pay-as-you-go), and GPU for startups (equity-based). Supporting the GPU infrastructure is the AIDC Platform Management Software for multi-brand GPU management, and value-added services including Model-as-a-Service and Data Pools. Trio AI operates Hong Kong's first Chinese high-performance 1024-GPU cluster using 400G IB network infrastructure, with plans to expand to over 10,000 GPUs. The platform integrates essential developer tools and DevOps capabilities to support AI model training and inference at enterprise scale.

Differentiator

Problem solved

Functional benefit

Products and services

• GPU-as-a-Service (GaaS) Cloud-based GPU resource access service enabling organizations to leverage high-performance computing power remotely. Provides scalable, adaptable, and cost-effective access to GPU resources for tasks requiring significant parallel processing power, backed by data centers in Hong Kong and Greater Bay Area.
• Model-as-a-Service Value-added AI platform service providing access to pre-trained models for development, enabling enterprises to implement AI applications without building models from scratch.
• Data Pools Value-added data service facilitating data collection, storage and application across Hong Kong, Shanghai and additional locations for cross-border clients and projects requiring compliant data transfer and connectivity.
• Training Service AI model training, fine-tuning and serving service that provides availability and economies of scale to empower users to train and deploy models at scale.
• Inference Service Scalable inference service that serves better inference and scales across GPUs as demand changes for production AI workloads.

Quantifiable outcome

• First 1,000-GPU cluster in Hong Kong launched August 2024
• +2 more outcomes

Named customers6 records

Segments4 records

Ideal customer profiles4 records

Technology focussed Yes

API detail

Has API

No

API docs

API detail

Core technology

AI maturity

App detail

AI capability2 records

Feature3 records

Partnerships

Seven partnerships are on record, tiered core and minor.

• HKT LimitedcoreChannel Partner/ Reseller/ Distributor · 11 December 2024Trio AI signed MOU with HKT on December 11, 2024 to explore opportunities in Artificial Intelligence Super Computing (AISC) solution through combined business networks. The partnership aims to resell, market and promote AISC solution in Hong Kong, Greater Bay Area, and Southeast Asia. HKT's AI Superhighway provides ultra-high-speed (800 Gbps) and ultra-low-latency (<1 millisecond) connectivity linking Trio AI's GPU cluster with enterprise data centers.
• City University of Hong Kong (CityUHK)coreStrategic or Co-development Partner · 21 October 2024Trio AI signed MOU with CityUHK on October 21, 2024 for strategic collaboration on GPU computing power to advance AI for Science. Partnership supports the Hong Kong Institute of AI for Science (HKAI-Sci), bringing together experts from computer science, data science, statistics, and bioinformatics to facilitate integration of science and AI.
• METAXcoreTechnology or IntegrationTrio AI collaborates with METAX to leverage AI computing resources supported by METAX chips, building Hong Kong's first Chinese high-performance 1,024-GPU cluster. The cluster uses 400G IB network + all-flash storage + high-performance GPU servers, optimized for various large models and efficient tuning with GPGPU architecture.
• Votee AIminorGTM or Marketing PartnerVotee AI, a leading generative AI platform, participates as Trio AI partner in panel discussions on digital transformation for Hong Kong customers. Represents ecosystem partner in AI application development.
• WintoominorGTM or Marketing PartnerWintoo is a national high-tech AI enterprise with cloud-based platform (雲上坦途) dedicated to leveraging AI for assessing urban road conditions. Partner in Trio AI ecosystem panel discussing Hong Kong digital transformation.
• YedpayminorGTM or Marketing PartnerYedpay partners with Trio AI to reshape payment landscape using AI with focus on security and user experience through Identity, Individualization, and Insight (3i) framework.
• Valuetronics Holdings Ltd.minorOthersExecutive Director Mr. Chow Kok Kit of Valuetronics Holdings Ltd. witnessed the MOU signing between Trio AI and HKT, indicating involvement in the partnership ecosystem.

Scale indicators7 records

Recent moves6 records

Expansion highlights6 records

Direct peers

• CoreWeave: CoreWeave is a US-based GPU cloud provider purpose-built for AI training and inference workloads, operating large-scale NVIDIA GPU clusters. It is the closest direct peer to Trio AI's GPU-as-a-Service model with multi-vendor GPU access, customized enterprise contracts, and on-demand self-service offerings.
• Nebius: Nebius is an AI infrastructure company operating full-stack GPU clusters and cloud services for AI training and inference globally. It is comparable to Trio AI's GPU-as-a-Service offering with multi-brand GPU management and enterprise-grade capacity.
• Crusoe: Crusoe provides large-scale GPU cloud infrastructure purpose-built for AI training and inference, leveraging stranded energy. It is comparable to Trio AI as a specialized GPU cloud provider serving enterprise AI customers with purpose-built compute capacity.
• Together AI: Together AI provides a GPU cloud platform optimized for open-source model training, fine-tuning, and inference at scale. It competes with Trio AI in offering managed AI compute with developer-oriented tooling, though at much larger scale.
• RunPod: RunPod offers GPU cloud services including on-demand and reserved GPU instances for AI training and inference. Like Trio AI, it targets both startups and enterprises with flexible, developer-friendly GPU access models.
• Lambda Labs: Lambda Labs operates GPU cloud infrastructure and on-demand clusters for AI training and inference, serving AI startups and enterprises. Like Trio AI, it offers hybrid reserved/on-demand GPU access and developer tooling tailored to AI workloads.

Regional players

• Hetzner: Hetzner is a European-based cloud provider offering GPU instances at competitive price points. It is comparable to Trio AI's value-oriented, developer-friendly GPU cloud positioning but serves a different geography.

Emerging players

• Vast.ai: Vast.ai operates a decentralized GPU marketplace connecting compute buyers with providers globally, including across Asia. It is comparable to Trio AI's heterogeneous GPU access model but uses a marketplace rather than owned/operated infrastructure.

Broad incumbents

• Alibaba Cloud (Aliyun): Alibaba Cloud is the dominant cloud and AI infrastructure provider in Asia, offering GPU instances for AI workloads across Greater Bay Area and globally. It competes broadly with Trio AI's GaaS offering but as a hyperscaler with much larger GPU fleets and broader service portfolio.
• Tencent Cloud: Tencent Cloud provides GPU-accelerated cloud services across Asia for AI training and inference, with strong presence in Hong Kong and Greater Bay Area. It overlaps with Trio AI's regional footprint but offers a much broader cloud platform.

Market position

Strengths5 records

Weaknesses5 records

Competitive moat5 records

Key risks6 records

Key highlights7 records

Customer concentration

• Facebookfacebook.com/profile.php
• LinkedInlinkedin.com/company/trioai

Revenue estimate

Valuation estimate

Number of profiles

Profiles5 records

Funding overview

Funding rounds

Investors

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

M&A

Investments

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

Frequently asked questions about Trio AI