NVIDIA Triton Inference Server for Linux suffers a resource exhaustion flaw that allows an attacker to trigger unbounded allocation of system resources. The failure can lead to denial of service by exhausting memory or CPU, preventing legitimate users from accessing the service. This weakness falls under CWE-770, the category of Resource Exhaustion.

Affected are NVIDIA Triton Inference Server installations running on Linux platforms. Specific affected versions are not enumerated in the available data.

The CVSS score of 7.5 signals a high severity vulnerability. EPSS data is unavailable, so the probability of exploitation is unknown, though the absence from the CISA KEV catalog suggests no confirmed public exploits yet. The likely attack vector is remote, via crafted inference requests sent to the server; a successful exploit would disrupt availability for all users without providing privilege escalation or data disclosure.