- Despite concerns about an AI bubble, Nvidia argued the tech actually isn't receiving enough attention
- Nvidia has found itself at the epicenter of telco efforts to embrace AI across all parts of their operations
- The vendor is set to unveil its latest advancements this week at its annual Nvidia GTC conference
For all the hoopla surrounding artificial intelligence, Nvidia SVP Ronnie Vasishta says the technology is "underhyped."
"We're three years into a new cycle and the opportunity for AI is it will touch every part of every industry," Vasishta said in an interview with Fierce at Mobile World Congress in early March.
AI was the primary focus of MWC — that much was obvious, Vasishta noted. "Now let me tell you what's not so obvious — three years ago, that wasn't t=he case. If you came to MWC three years ago, the first show after the pandemic, AI wasn't much of a factor." The main focus for MWC three years ago was the metaverse and edge applications.
This year, though, it seemed like every booth and stand at the conference had an AI label slapped prominently on it.
But AI isn't just one thing; there are a variety of use cases for telcos: Network operations, internal productivity, customer experience and call centers, where AI helps customers with support issues such as changing plans, changing operators and identifying network faults that disrupt service.
And an ecosystem of vendors is building for each use case, comprising startups and established vendors, Vasishta said.
But it's not exactly clear how many of the use cases showcased at MWC will be around for the long haul. Gartner recently predicted that by the end of 2025, nearly a third (30%) of all generative AI use cases currently in the proof of concept stage will be scrapped due to escalating costs, unclear business value, poor data quality or inadequate risk controls.
Gartner has also indicated that AI has passed the peak of inflated expectations and will head into the trough of disillusionment in the coming months and years. That said, the firm noted AI will continue to shape industries across the board and tipped the next frontier beyond generative AI to be composite AI, which combines different AI methodologies to create more powerful and scalable AI solutions.
The telco AI promise
In the telco realm, some operators are using AI to improve productivity, while others use AI to improve network operations, reducing OpEx by automating network operations. Nvidia has found itself at the epicenter of many of these efforts.
For instance, SoftBank and T-Mobile are working with Nvidia to move to an infrastructure that can support both RAN and AI. Nvidia is also partnering with T-Mobile, SoftBank, Ericsson and Nokia on software-defined RAN for mobile switching offices.
Elsewhere, Nvidia is working with Indosat Ooredoo Hutchison on building an AI factory — a data center specialized for AI processing — using Nvidia reference architecture. Some 14 companies worldwide have built AI factories with Nvidia, Vasishta said.
For more on the value of AI factories, read our Fierce Network Research report: "AI sovereignty: Seizing the opportunity for transformative telco infrastructure investment."
And Nvidia is partnering with Verizon on its AI Connect initiative to deliver on skyrocketing enterprise requirements for bandwidth and connectivity to support AI.
For telcos seeking to enhance customer experience, Nvidia can support automatic speech recognition and help build "digital humans" with realistic faces.
To implement AI use cases, Nvidia works with all the top telcos in the world, Vasishta said. The company doesn't just provide processors; it delivers "full stack building blocks" — both hardware and software — to the ecosystem, which builds solutions based on those blocks.
Nvidia's AI software includes its CUDA platform for parallel computing and APIs; Nvidia AI Enterprise, for developing enterprise AI applications; and vertical AI application libraries for healthcare, retail, industrial, and telecoms. Its partners include companies such as Amdocs, ServiceNow and Salesforce. Applications include service ticket resolution, OSS billing, customer service, customer experience, and tools for resolving trouble tickets more quickly.
Additionally, Nvidia is a founding member of the AI RAN Alliance, established a year ago at MWC with 10 members, including T-Mobile and SoftBank. That organization is now up to more than 75 members, Vasishta noted.
It's clear there's momentum behind AI in the telco space. But as Gartner noted there are still risks that at least some deployments won't pan out as expected. Why? "Even when a business safely deploys an AI model and achieves regulatory compliance, scalability, technical and organizational complexity, low user acceptance and adoption and unforeseen difficulties can hinder enterprise wide AI implementation," the firm wrote.
So, what's Nvidia's plan to address these risks? And what new innovations does it have up its sleeve? We'll hear more from Nvidia at its big GTC AI Conference, taking place this week in San Jose, California. Notably, the conference includes a telco day, with programming on how AI can be used in the RAN to improve spectral efficiency.
Fierce Network will be on the ground, soaking up the news with a critical ear and bringing you highlights and analysis from the event. Stay tuned!