GIGABYTE Expands Its AI TOP Lineup With Powerful Desktop Workstations for Local AI and LLM Workloads
GIGABYTE has introduced a new family of AI-focused desktop workstations under its AI TOP ecosystem, aiming to give creators, developers, researchers, and AI enthusiasts ready-made systems for running demanding workloads locally. Revealed during the company’s ENTER INFINITY press event, the lineup includes three workstation-class machines designed for different levels of AI performance, from high-end desktop AI computing to serious workstation-grade processing.
The new systems include the AI TOP 100 B850, AI TOP 100 Z890, and AI TOP 500 TRX50. Each model is built around premium components, large memory configurations, powerful graphics options, and a 1600W power supply designed to handle heavy AI training and inference tasks.
The AI TOP 100 B850 is positioned as a high-performance AMD-based AI workstation. It appears to use GIGABYTE’s B850 AI TOP motherboard and combines AMD’s Ryzen 9 9950X processor with 128GB of DDR5 memory. For graphics, buyers can choose between an NVIDIA GeForce RTX 5090 32GB or an AMD Radeon AI PRO R9700 32GB.
This configuration gives the system enough GPU memory to handle large AI models, with GIGABYTE claiming support for large language models with up to 405 billion parameters. That makes the AI TOP 100 B850 a compelling option for users who want to run advanced LLMs locally rather than relying entirely on cloud-based AI platforms.
Power delivery comes from the UD1600PM PG5 AI TOP PSU, a 1600W 80 Plus Platinum certified power supply. This is important for a workstation built around power-hungry flagship GPUs and a high-end desktop CPU, especially when running sustained AI workloads.
GIGABYTE also says the system has been validated with more than 100 AI applications, though the company has not provided a detailed list of which software tools were tested. Still, this suggests the workstation is being positioned as a plug-and-play solution for AI development, model training, inference, data science, and content creation workflows.
The company also shared performance comparisons against what it described as a normal system, claiming up to 10 times better performance in epoch count and up to 237 times faster time-to-completion in certain workloads. While those numbers sound impressive, the lack of detailed testing conditions makes it difficult to judge the exact real-world advantage. Even so, the hardware inside these machines is clearly built for serious acceleration.
One of the more interesting choices in the AI TOP 100 B850 is the option between the RTX 5090 and Radeon AI PRO R9700. Both cards offer 32GB of VRAM, but they target different types of buyers. The Radeon AI PRO R9700 may appeal to users looking for strong AI memory capacity at a lower cost, while the RTX 5090 is expected to deliver faster performance thanks to its wider 512-bit GDDR7 memory bus and higher memory bandwidth.
The AI TOP 100 Z890 takes a different approach by using Intel’s platform. This workstation is built around the GIGABYTE Z890 AI TOP motherboard and features an Intel Core Ultra 9 285K processor, 128GB of DDR5 memory, and a GIGABYTE GeForce RTX 5090 Windforce OC 32GB graphics card.
Like the B850 model, the Z890 version uses the same UD1600PM PG5 AI TOP 1600W power supply, giving it plenty of headroom for demanding AI and GPU-heavy tasks.
The biggest highlight of the AI TOP 100 Z890 is connectivity. GIGABYTE emphasizes the system’s dual Thunderbolt 5 ports, which can deliver up to 80Gbps of bandwidth. According to the company, multiple workstations can be connected through Thunderbolt 5 to improve training performance, with claimed gains of up to 1.6 times in certain AI workloads.
That feature could make the AI TOP 100 Z890 especially attractive for small teams, labs, creators, or developers who want to scale AI training performance without moving immediately to a full server or cloud-based cluster. The ability to link multiple desktop workstations could offer a more flexible path for users building local AI infrastructure.
At the top of the lineup sits the AI TOP 500 TRX50, a much more powerful workstation built for users who need extreme CPU performance and massive memory capacity. This model uses the TRX50 AI TOP sTR5 motherboard and is powered by AMD’s Threadripper PRO 7965WX, a 24-core, 48-thread workstation processor.
The system also includes a huge 768GB of DDR5 memory, making it suitable for memory-intensive AI workloads, large datasets, simulation, rendering, engineering applications, and professional workstation tasks. With this amount of RAM, the AI TOP 500 TRX50 can even run some large language models directly on the CPU, although the included RTX 5090 gives users a much faster GPU-based option for many AI tasks.
Once again, the system is powered by GIGABYTE’s UD1600PM PG5 AI TOP 1600W power supply, keeping the lineup consistent across all three models.
One unusual detail about the AI TOP 500 TRX50 is its apparent limitation around multi-GPU expansion. Despite being based on a workstation platform that is typically well-suited for multiple graphics cards, the system does not seem to support an additional GPU in full x16 mode. This may be due to the large 3.5-slot RTX 5090 used in the system.
That design choice may feel limiting for users who expected the TRX50 platform to focus heavily on multi-GPU setups. However, GIGABYTE appears to be positioning this machine more around CPU strength, memory capacity, and high-end single-GPU AI acceleration rather than maximum GPU expansion.
Overall, GIGABYTE’s AI TOP workstation lineup gives the company a stronger presence in the growing market for local AI computing. As more users look for ways to train, fine-tune, and run AI models on their own hardware, prebuilt AI workstations like these could become increasingly appealing.
The AI TOP 100 B850 offers a strong AMD Ryzen-based platform with flexible GPU options. The AI TOP 100 Z890 focuses on Intel performance and Thunderbolt 5 connectivity for workstation linking. The AI TOP 500 TRX50 delivers workstation-class CPU power and massive memory capacity for more demanding professional workloads.
With configurations built around the RTX 5090, Radeon AI PRO R9700, Ryzen 9 9950X, Core Ultra 9 285K, and Threadripper PRO 7965WX, GIGABYTE is clearly targeting users who want serious local AI performance without having to assemble a custom workstation from scratch. For developers, AI researchers, creators, and professionals working with large language models, these systems could offer a powerful alternative to cloud-only AI workflows.
