Nvidia Survey Reveals How Telcos Plan to Use AI; Quantifying ROI is a Challenge
A Nvidia sponsored survey of more than 400 telecommunications industry professionals from around the world found a cautious tone in how they plan to define and execute on their AI strategies. Virtually every telco is already engaged with AI in some way, although mostly at an early stage. NVIDIA’s first “State of AI in Telecommunications” survey consisted of questions covering a range of AI topics, infrastructure spending, top use cases, biggest challenges and deployment models. The survey was conducted over eight weeks between mid-November 2022 and mid-January 2023.
Amid skepticism about the money-making potential of 5G, telecoms see efficiencies driven by AI as the most likely path for returns on investment. 93% of those responding to questions about undertaking AI projects at their own companies appear to be substantially underinvesting in AI as a percentage of annual capital spending.
Some 50% of respondents reported spending less than $1 million last year on AI projects; a year earlier, 60% of respondents said they spent less than $1 million on AI. Just 3% of respondents spent over $50 million on AI in 2022.
The reasons cited for such cautious spending? Some 44% of respondents reported an inability to adequately quantify return on investment, which illustrates a mismatch between aspirations and the reality in introducing AI-driven solutions. 34% cited an insufficient number of data scientists as the second-biggest challenge.
The biggest telco objectives for AI are to: optimize operations (60%), lower costs (44%) and enhance customer engagement (35%). Respondents cited use cases ranging from cell site planning and truck-route optimization to recommendation engines.
Just over a third of respondents said they had been using AI for more than six months. 31% said they’re still weighing different options, 18% reported being still in a trial phase and only 5% said they had no AI plans at all. Most industry execs say they see AI technologies will positively impact their business – 65% agreed AI was important to their company’s success, and 59% said it would become a source of competitive advantage.
Operators are spending a fraction of their capex budgets on AI projects – last year half said they spent less than $1 million on AI. At the top end, 2% spent more than $50 million in 2021, with that number rising to 3% in 2022.
The latest AI Index compiled by Stanford University puts telcos at the forefront of AI deployment. Using its own data and that from a McKinsey study, it found that the highest level of AI adoption is in product or service development by hi-tech companies and telcos (45%), followed by AI in service operations (45%).
The biggest single application in any industry was natural language text understanding deployed by 34% of hi-tech and telco firms, with 28% implementing AI-based computer vision and 25% using virtual agents.
- Moving from proof of concept to production/scale 47%
- Economic uncertainty 46%
- Infrasctructure upgrades 46%
- Market differentiation 34%
- Change in priority of data science 20%
- 92% will either increase or maintain their AI spend in 2023.
Allied Market Research: Global AI in telecom market forecast to reach $38.8 by 2031 with CAGR of 41.4% (from 2022 to 2031)
Global AI in Telecommunication Market at CAGR ~ 40% through 2026 – 2027
The case for and against AI in telecommunications; record quarter for AI venture funding and M&A deals
SK Telecom inspects cell towers for safety using drones and AI
Summary of ITU-R Workshop on “IMT for 2030 and beyond” (aka “6G”)
One thought on “Nvidia Survey Reveals How Telcos Plan to Use AI; Quantifying ROI is a Challenge”
SKT is arguably more invested in AI than any other telco worldwide. Back in November, company executives announced plans that put AI at the very heart of SKT’s future strategy. “We will transform our core businesses into AI-based businesses to secure new growth drivers, offer AI services to deepen customer relations and employ AIX strategy to spread SK Telecom’s AI and digital transformation capabilities to other industries,” said Jinwon Kim, the operator’s chief financial officer.
Evidence of this approach could be seen at MWC, where SKT was eager to parade some of its AI partners and innovation. “We’re accelerating AI transformation with Korean AI companies,” said Jang Seong-ho, SKT’s AI business development team leader, during a conversation with Light Reading. “Together with partners, we’re trying to build an AI ecosystem in the global market, including the US.”
Phantom AI is one such partner. Founded in 2016, and based in Silicon Valley, it is developing algorithms for use in so-called advanced driver assistance systems (ADAS), where vehicles would be only partly autonomous. Industry people refer to five levels of vehicle autonomy, with 1 meaning basic driver assistance and 5 describing a car that takes humans entirely out of the driving process – allowing the person who sits behind the wheel to knit, read novels or sleep.
“We are not targeting level 4 or 5 full autonomy but focusing on levels 2 and 3,” said Cho Hyung-gi, Phantom AI’s founder and CEO, who met with Light Reading at SKT’s stand. His company now has a contract with a major vehicle manufacturer and hopes to be in production by the final quarter of this year.
SKT’s practical involvement with Phantom AI seems to lie on the semiconductor side through an affiliate called Sapeon. Like Phantom, it is headquartered in California but wholly owned by SK Group, the parent company of SKT and South Korea’s second-largest “chaebol,” or conglomerate, after Samsung. Viewed as a potential rival to Nvidia, Sapeon is currently working on AI chips that could be used for servers and edge devices as well as in ADAS, said Jang.
“That is the idea,” Cho told Light Reading. “The subsidiary is well-known, and they are making an innovative AI chipset and we are purely software and so we see the synergy between Phantom and Sapeon.” A low-cost system-on-a-chip (SoC) priced at between $20 and $30 could be vital in helping Phantom target a mass market of lower-priced vehicles, Cho explained. “Tesla is using a very powerful and expensive SoC, but we are running on a low-cost chip so that we can deploy to millions.”
Microsoft was talking up a version of GitHub Copilot – a software-writing AI based on the same GPT-3 platform as ChatGPT – tailored for telco needs. A few telcos are examining how ChatGPT could be used in operations for things such as software documentation and to help field workers and customer service agents.
Howard Watson, the chief security and networks officer of the UK’s BT, even spies a potential sales opportunity. “The thing about ChatGPT is the immense compute that’s required, particularly in the learning phase, and one of the great ways of distributing that – and why I’m thinking is generative AI an opportunity – is between the edge and the devices,” he said. “Could it be the killer app for multipurpose edge?” BT’s shareholders will certainly hope so.