Nvidia RTX 5090: Peak Power Consumption and How to Manage It

The Nvidia GeForce RTX 5090, launched in January 2025 for $1,999, stands as the most powerful consumer GPU available. Its impressive performance comes with a significant power draw, officially rated at 575 watts. Real-world testing by Overclocking on a Founders Edition card revealed an average consumption of 559 watts during 4K gameplay in “Cyberpunk 2077. ” More notably, the RTX 5090 experiences momentary “power excursions,” brief spikes that far exceed its official rating. Igor’s Lab recorded spikes reaching 738 watts for up to 10 milliseconds and even 901. 1 watts for durations under a millisecond. While these figures might seem alarming, they are typical for high-end graphics cards. These transient loads are compatible with the ATX 3. 1 power supply specification. However, older PSUs adhering to the ATX 2. 51 standard may struggle, as this older specification lacks official guidelines for handling such rapid power fluctuations. Older power supply units might misinterpret these spikes as faults, triggering overcurrent protection and shutting down the system. Conversely, ATX 3. 0 and 3. 1 standards were specifically designed to accommodate these demanding transient loads. For users concerned about power consumption, undervolting offers a viable solution. One user reported maintaining 99% of the RTX 5090’s stock performance while reducing power usage by 16%, lowering the average draw to 464 watts. Aggressive undervolting even led to improved 3DMark scores by creating thermal headroom. This allows the GPU’s boost algorithm to maintain higher clock speeds, achieving better overall performance. Nvidia recommends a 1000W power supply unit for this powerful GPU.