Nvidia Corp. today introduced Morpheus, a software framework that will allow enterprises to detect breaches more effectively by allowing them to scan a bigger portion of their internal network traffic for threats.

The framework uses a combination of artificial intelligence and data processing unit chips to provide the increased visibility. It debuted today at Nvidia’s GPU Technology Conference virtual event alongside BlueField-3, Nvidia’s newest data processing unit, which integrates with Morpheus. 

Modern breach prevention tools work by scanning the traffic that goes through a company’s network for unusual patterns. The corporate network is the medium through which malware spreads from server to server, which makes it an invaluable source of security information. But because of technical limitations, breach prevention tools can usually only analyze so much of that security information. Nvidia estimates that traditional cybersecurity tools examine a mere 5% of the data traffic going through corporate networks. 

With the combination of its new Morpheus framework and BlueField-branded DPUs, Nvidia says that it can overcome that limitation. BlueField chips can be attached to servers to provide extra computing power for certain applications. Morpheus harnesses that extra computing power to let enterprises analyze every single packet going through the network instead of just a small portion. The result should, in theory, be faster detection of threats inside corporate data centers. 

The BlueField chips have optimizations that allow them to perform certain tasks far more efficiently than severs’ built-in central processing units. Scanning network traffic is one of those tasks. The newest chip, the BlueField-3 DPU announced today, can provide performance equivalent to 300 CPU cores in some scenarios. 

Because cybersecurity products typically use AI under the hood to look for threats, Morpheus also provides an array of machine learning features. The main highlight: a selection of pre-trained AI cybersecurity models. They can spot ransomware, encryption keys being transported by hackers across the network and even catch certain types of server malfunctions, among other types of issues.

With its ability to boost the capabilities of breach prevention tools, Morpheus should be of particular interest to cybersecurity software companies. One such companies, Splunk Inc., is already working with Nvidia to explore potential uses for the framework.

Morpheus could give Nvidia’s DPU business a boost by creating a compelling security use case for the chips and thereby give enterprises another reason to buy them. In the future, Nvidia might seek to develop additional software tools for its DPUs to unlock additional market opportunities. 

DPUs such as Nvidia’s BlueField series are not designed solely for scanning network traffic. Rather, chips in the category can perform a variety of low-level processing tasks that are normally relegated to server CPUs. The idea is that, by offloading certain tasks from server CPUs, more computing resources can be made available for business applications.

Nvidia said last year that it was aiming to introduce the next iteration of its DPU, BlueField-4, in 2023 with targeted AI performance of 400 trillion operations per second. As the chips’ capabilities continue to advance, they should become an even more competitive choice for use cases such as threat detection.

Image: Nvidia