AI Data Centers in a Box: Exascale and Compal Debut Integrated Solution

SAN FRANCISCO and TAIPEI, Taiwan – May 19, 2026 – As the artificial intelligence boom strains global power grids and data center capacity, a new partnership is aiming to deliver a complete, deployment-ready AI data center in a fraction of the traditional time. AI infrastructure startup Exascale Labs Inc. and global IT manufacturing giant Compal Electronics, Inc. have announced they will jointly showcase a fully integrated AI infrastructure solution at the upcoming COMPUTEX Taipei 2026 conference.

The collaboration promises to fuse four critical layers—compute, cooling, infrastructure, and power—into a single, factory-built offering. This move directly confronts the primary bottlenecks hindering the rapid expansion of AI: immense power consumption, extreme heat generation, and lengthy deployment timelines for new facilities.

The AI Infrastructure Bottleneck

The demand for AI is growing at an explosive rate, with the market for AI in data centers projected to surge from $6.5 billion in 2023 to over $73 billion by 2030. This exponential growth, however, comes with monumental challenges. Training and running large AI models require dense clusters of powerful GPUs, which consume vast amounts of electricity and generate unprecedented levels of heat. A single high-end AI server rack can now draw over 100 kilowatts of power, an order of magnitude greater than traditional IT racks.

This has pushed conventional data center designs to their breaking point. Traditional air-cooling methods are proving insufficient and inefficient for dissipating such concentrated heat, leading to performance throttling and increased energy waste. Industry analysts predict that over half of all data centers will need to adopt some form of liquid cooling by 2028 to cope with high-density workloads. Furthermore, the construction of a traditional data center is a multi-year endeavor, creating a significant lag between the demand for AI compute and its availability.

A Four-Layer Integrated Solution

The Exascale and Compal partnership aims to solve this complex puzzle by integrating their respective technological strengths into a cohesive stack. The joint solution, set to be demonstrated at Compal's booth during COMPUTEX, combines Compal's hardware expertise with Exascale's innovative infrastructure architecture.

At the base of the stack are Compal's latest AI server platforms, including the OG231-2-L1, OG430-2-L1, and SGX30-2 systems. These servers are optimized for the demanding tasks of AI model training and inference. To manage the intense heat produced by the GPUs within these servers, Compal is integrating its Direct Liquid Cooling (DLC) technology. DLC circulates fluid directly over the processors, transferring heat far more efficiently than air. This approach not only prevents thermal throttling, boosting performance and reliability, but also dramatically lowers the energy required for cooling, contributing to a much lower Power Usage Effectiveness (PUE) and reducing operational costs.

Exascale Labs provides the other half of the equation: the physical infrastructure and power system. The company's Modular Data Center (MDC) is a pre-fabricated, containerized unit where the servers, cooling, networking, and power systems are all pre-integrated and tested at the factory. This 'data center in a box' approach is central to the partnership's promise of rapid deployment.

Powering this high-density environment is Exascale’s novel HVDC / 800V DC power architecture, which utilizes Solid-State Transformers (SST). This advanced system eliminates multiple inefficient AC-to-DC power conversion steps found in traditional data center power chains. The result is a 10-15% reduction in energy loss, a simpler and more reliable power path, and a significantly smaller physical footprint for electrical equipment, freeing up valuable floor space for more revenue-generating GPU racks.

From Years to Months: The Promise of Rapid Deployment

The most compelling proposition of this partnership is the dramatic acceleration of deployment timelines. By delivering a pre-integrated, pre-tested solution, the companies claim they can slash the time from a purchase order to live AI capacity from years down to a matter of months.

“By pairing Compal’s AI servers and direct liquid cooling with our MDC and HVDC/SST stack, joint customers can move from purchase order to live AI capacity in months rather than years — without compromising on density or efficiency,” said Dr. Hoansoo Lee, CEO of Exascale Labs, in the official announcement. “COMPUTEX is the right stage to show the industry what this full pairing looks like in a single, deployable solution.”

This speed-to-market is a critical advantage for enterprises racing to build out their AI capabilities. Instead of navigating a complex web of vendors for servers, cooling, power, and construction, customers can acquire a complete, optimized system from a single point of contact.

“AI workloads have pushed rack-level power and thermal requirements into territory that demands fresh thinking on both the server and the facility,” added Alan Chang, Vice President, ISBG at Compal. “Our AI server platforms and DLC address the chip- and rack-level challenge; Exascale’s modular data center and HVDC architecture address the building- and grid-level challenge.”

Strategic Moves in a High-Stakes Market

This collaboration represents a significant strategic maneuver for both companies in the fiercely competitive AI hardware market, which is currently dominated by giants like NVIDIA, Dell, and HPE, alongside other major ODMs like Supermicro and QCT. For Compal, a manufacturing powerhouse known for notebooks and smart devices, this partnership marks a decisive step up the value chain. It positions the company not just as a manufacturer of server components, but as a provider of sophisticated, integrated AI solutions.

For Exascale Labs, the timing is crucial. The company is in the process of going public through a business combination with the SPAC D. Boral ARC Acquisition I Corp. (BCAR). The deal, expected to close in the second half of 2026, will see the combined company list on the Nasdaq exchange under the ticker symbol XLAB. The high-profile showcase at COMPUTEX with an established partner like Compal serves as a powerful demonstration of its technology and business model to potential investors ahead of its public debut.

By combining best-of-breed components into a single, deployment-ready stack, Exascale and Compal are betting that efficiency and speed will be the winning formula. As enterprises and researchers continue to push the boundaries of artificial intelligence, the underlying infrastructure must evolve to keep pace, and this integrated approach offers a compelling glimpse into the future of how AI will be powered.