- Juniper Networks showcased its AI-native networking platform tailored for "neocloud providers"
- "Neoclouds" are a recent buzzword describing providers specializing in GPU-as-a-service and AI-as-a-service
- Juniper also introduced a new branch switch and hardened metro router
MOBILE WORLD CONGRESS, BARCELONA — We first heard the buzzword "neocloud" Monday evening, and by Tuesday morning, Juniper Networks was telling us about new products to fit the market.
Travis Ewert, global head of platform development at Digital Realty, mentioned the neocloud in passing at a Cisco reception Monday night. We buttonholed him afterward to get an explanation (this is how we phrased the question: "Huh?"), and he explained this was a new generation of cloud providers, including CoreWeave and Vultr, focused on GPU-as-a-Service and AI-as-a-Service (GPUaaS and AIaaS).
The next morning, a quick Internet search uncovered several articles on the subject dating back to October, including one with a nifty image of Keanu Reeves as Neo from "The Matrix." Uptime Institute names a half-dozen players in the market, including startups CoreWeave, Paperspace, Nebius, Lambda Labs and Vast AI, as well as established players including OVH, Vultr, DigitalOcean and Rackspace.
Although the buzzword was new to us, we've written about the trend. We identified "AI factories" — data centers specialized for AI training — as a telco opportunity in a recent, free report: "AI sovereignty: Seizing the opportunity for transformative telco infrastructure investment". And we've talked about the companies, including CoreWeave and Vultr. Also, we've touched on GPUaaS in our coverage of GPUs.
Still, we love a good buzzword here at Fierce Network, and we were delighted to hear it again at a Tuesday morning briefing with Juniper.
Last week, the vendor introduced a solution purpose-built for neocloud providers, traditional service providers and other AI cloud providers who are deploying and managing GPUaaS and AIaaS to speed up deployments and simplify operations.
Juniper offered conversational AI before it became cool
Juniper also discussed its AI Native Networking Platform, in development more than a decade, which optimizes networks using a combination of telemetry, microservices and AIops, and which has been using a conversational interface for nine years, according to Jeff Aaron, Juniper GVP Product Marketing. This was well before generative AI became commonplace with the launch of ChatGPT into public consciousness. Juniper also showcases its AI networking, including switches, routers and data center interconnect.
Juniper recently beefed up its AIops capabilities with improved edge routers and a new branch switch—the EX4000. This smaller, noiseless switch with no fans and a boot time under two minutes is designed for clinics, retail sites and small branch offices where quiet and fast boot time matter. Additionally, Arista introduced its ACX7020 Cloud Metro Router, a hardened platform offering 100 gigabits throughput that can be deployed indoors or outdoors.
Juniper's latest steps come as it waits for the resolution of HPE's $14 billion acquisition bid, which the U.S. Department of Justice has blocked. Aaron said, "It's got to play out as a process. You know, obviously, we still very vehemently believe that the DOJ was wrong. I think a lot of analysts agree with us, and we think it's just going to take longer now. We still think it's gonna come through.”
Juniper faces tough competition in AI networking from the likes of Cisco, Arista, Extreme Networks and Dell Technologies. In particular, Cisco recently introduced its N9300 Series Smart Switches, integrating Hypershield security to simplify AI data center operations, going up against Juniper's product line.
Stay up to date on all the news from Mobile World Congress with Fierce Network and FNTV.