Trending Articles
Trending Topics
Trending
There’s one hot message coming from MWC Barcelona 2025: the future of mobile connectivity is hurtling towards us. A future powered by 6G and AI. The problem for telcos such as Verizon, T-Mobile and Orange is that their hardware isn’t tuned for these accelerated demands. And that, says HPE, is where the ProLiant Compute DL384 Gen12 server steps in.
As Gerasimos Klaoudatos, Global Sales Lead at HPE Aruba Networking, emphasised when I talked to him on the HPE stand, the HPE ProLiant Compute DL384 Gen12 does not stand alone. It also introduces an architecture, based on Nvidia’s Blackwell superchip, that HPE will expand upon.
Who, other than telcos, will need this product and its relations? During our conversation, Gerasimos mentioned examples such as universities and manufacturing companies who want to run secure AI apps on a private 6G cloud. It’s right for you, he said, “if you have to run your own code, AI solutions, things like that”.
And because it supports multiple tenants, such a piece of hardware could be run by a managed service provider for a variety of clients.
HPE introduced the ProLiant Compute DL384 Gen12 server six months ago, following it with the launch of its expanded ProLiant Compute Gen 12 portfolio last month.
These all uses Intel’s latest Xeon 6 processors, and crucially they include the Nvidia Blackwell H200 “superchip”. The DL384 is a 2U rack server that supports up to two Intel Xeon 6 processors, and for obvious reasons it’s geared towards AI acceleration.
“We’ve already scanned the market, we already understand what the market wants, especially when it comes to 6G,” said Klaoudatos. It’s at this moment, he explains, that the DL384 – and the wider AI accelerator platform it forms part of – will be absolutely necessary.
This doesn’t mean that telcos can wait for five years. Klaoudatos explained that they need to prepare themselves for the future so that they have already optimised their applications and can introduce this new generation of hardware seamlessly.
The crucial word there is “optimise”. It’s not enough for the apps to run: they must do so quickly and with as low energy demands as possible. “At this point in time, we’re working to optimise the power consumption, we’re working out how to put more workloads on Nvidia’s superchip architecture,” said Klaoudatos.
There’s also the simple matter of keeping users connected, he added, and that will again require AI inferencing directly on the server because 6G is far more complicated than 5G.
What’s more, it needs to happen all the time and for all users. “People are always moving, so we need to find out which is the best channel [to connect on],” said Klaoudatos.
“Who can predict when you move? You may be sleeping at night, but then in the middle of the night, you may decide to get in the car and go a trip and then you are not going to use the same channel. So that inference is happening all the time.”
