In this video, Alex embarks on building a powerful AI rig featuring a Threadripper 9970X CPU and two RTX Pro 6000 GPUs, which demands an enormous amount of power. Initially, his all-in-one AI rig was compact, but the addition of a second RTX Pro 6000 made it clear that a bigger setup was necessary. To accommodate the high power requirements, Alex visits MicroEnter, a computer parts store, to consult with his friend Dan about suitable power supplies, cases, and cooling solutions for this monster build.

The power demands of the system are immense, leading to a discussion about whether to use a single massive 2500W power supply or split the load between two smaller power supplies. Due to limitations in typical office wiring and the need for special high-voltage plugs, the dual PSU setup seems more practical. Dan and Alex explore various cases, ultimately settling on a large full tower case capable of housing the massive components and power supplies. They also consider the aesthetics and practicality of the case, including limited edition designs and features like panoramic OLED screens.

Power supply efficiency is a key concern, with titanium-rated units preferred for their up to 95% efficiency, which minimizes wasted energy as heat. Dan recommends a combination of a 1650W power supply for the CPU and one GPU, paired with a smaller SFXL power supply for the second GPU. They also discuss synchronization adapters to ensure both power supplies work in tandem. Additionally, Alex plans to use an uninterruptible power supply (UPS) to protect the rig from power fluctuations, though these units are expensive and heavy.

Cooling is another critical aspect of the build, with Alex opting for high-quality, quiet Noctua fans to keep the system cool without excessive noise. The case design allows for easy installation of fan assemblies, making maintenance simpler. After assembling the components and installing drivers, Alex tests the rig with a massive 235 billion parameter AI model called Quen 3, which requires the dual RTX Pro 6000 GPUs to run effectively. The model runs smoothly, utilizing both GPUs efficiently and demonstrating impressive performance.

In conclusion, Alex successfully builds a high-powered AI workstation capable of handling extremely large models that typical gaming GPUs cannot manage. The video highlights the challenges of powering and cooling such a system, the importance of choosing the right components, and the satisfaction of seeing the rig in action. Alex thanks MicroEnter for their support and invites viewers to check out their new store, promising more updates and in-depth testing of the RTX Pro 6000 in future videos.