Meta, Alphabet, Amazon, and Microsoft mentioned AI over 150 times during their Q1 earnings calls, and switchmakers like Arista Networks are saying “aye-aye” in response.
Massive AI investments require massive data centers, and those data centers call for networking components that send communications between server-rack stacks, or clusters. A switch provides the set of roads for data to travel as users request it from the data center. Companies that supply this kind of data-center network infrastructure are poised to profit off of an incipient AI boom.
“Presently, we are in the midst of trials leading to production deployments this year in 2023. We expect AI networking to become meaningful throughout the years and through the decade ahead,” said Arista Networks CEO and President Jayshree Ullal on a May 1 earnings call.
And now, the AI-nvestments…
- Google introduced its language-learning model Bard in March and plans to continually incorporate generative AI into its search function.
- Microsoft’s GitHub Copilot is now powered by OpenAI’s new GPT-4 model. The company also recently announced an AI-powered Bing and Edge, as well as a Bing Image Creator, which builds an image from one’s words.
- Amazon is working on its own language-learning model and new tools for building with generative AI on AWS.
- Meta has used the AI models to support the recommendation systems for ads and news feeds, and even wants to use AI to generate advertisements.
Meta said its Q1 capital expenditures of $7.1 billion were “driven by investments in data centers, servers, and network infrastructure.”
Top insights for IT pros
From cybersecurity and big data to cloud computing, IT Brew covers the latest trends shaping business tech in our 4x weekly newsletter, virtual events with industry experts, and digital guides.
Arista Networks, which announced Q1 revenues that were up 54% year over year, is used by both Microsoft and Meta.
A forecast from International Data Corporation (IDC) predicted that global spending on AI will reach $154 billion in 2023 and $300 billion or more by 2026. The 2023 figure represents an increase of 27% compared to 2022.
“Data center spending has been pretty stable, and networking spending specifically, has been stable or growing,” Brandon Hoff, research director at the market-intelligence firm IDC, told IT Brew, adding that the strength of the market is in AI and machine-learning processing.
Data-center switch companies like Nvidia (through its Mellanox acquisition), Arista, and Cisco have an optimistic outlook, Hoff told IT Brew. Doubling network speeds through switches means more efficient use of expensive hardware like GPUs (with Nvidia being one of the largest GPU manufacturers).
“Just a small investment in switching can save billions of dollars in terms of overall hardware,” said Hoff, who sees a growth in “high-end, high-performance” workloads as both the hyperscalers and large enterprise customers try out AI.
Cisco Systems CEO Chuck Robbins, sees major promise with AI.
“These AI networks that are being built…they’re bigger than the core-infrastructure networks that they’re running, which was astonishing to me when I learned that. And the network performance required is three to four times what they’ve historically needed,” Robbins said on an earnings call in February.
“And so, this is a massive opportunity for us,” he added.
You can say that again.