Streams in Node.js are designed for handling large amounts of data efficiently, but if backpressure isn't handled properly, you can end up with high CPU usage. Backpressure occurs when the writable side of a stream can't keep up with the readable side, leading to the application consuming too much CPU to process incoming data. You can control this by listening for the `drain` event in writable streams and using `pause()` and `resume()` on readable streams to slow down the flow of data when necessary. Profiling tools can help you pinpoint where the bottleneck occurs and adjust the data flow logic to balance the stream consumption rate. Properly managing backpressure ensures that your streams perform optimally under load without hogging CPU resources.