Mark your calendar for Mediaweek, October 29-30 in New York City. We’ll unpack the biggest shifts shaping the future of media—from tv to retail media to tech—and how marketers can prep to stay ahead. Register with early-bird rates before sale ends!
Tech industry leaders from Google, Salesforce, Adobe, Anthropic, CoreWeave and Amazon Web Services met at Bloomberg Intelligence’s annual AI summit at Bloomberg’s New York headquarters to discuss advancements in generative artificial intelligence, like the rise of autonomous agents and the soaring demand for data centers.
Enterprise spending on gen AI is expected to reach an eye-popping $151.1 billion by 2027, per eMarketer, and advancements in the technology are progressing at a breakneck pace, spurring tech and marketing leaders to debate topics like large-language models for emerging modalities, including video integration, as well as copyright risks in text-to-video tools.
Here are some of the key takeaways for marketers.
AI use cases are only just emerging
Neerav Kingsland, head of global accounts at Anthropic, which snagged a $2.75 billion investment from Amazon last week, said tangible enterprise applications are only just beginning: LexisNexis is using Anthropic’s LLMs on Amazon Bedrock for legal analysis.
Anthropic’s suite of LLMs, Claude, debuted in March 2023.
“By April, we had Fortune 100 companies knocking on our door,” said Kingsland. “But it was all proof of concept. Nothing was really getting deployed in 2023.”
Brian Venturo, co-founder and chief strategy officer at Nvidia-backed CoreWeave, said consumer LLMs are currently in a “novelty phase,” and it will take approximately five years to fully understand its implications for companies.
“Nobody really knows how this is going to permeate our daily lives,” he adde.
Demand for data centers is grossly underestimated
Venturo said the daily data center requests he receives are “absurd,” with some clients seeking entire campuses exclusively for their use.
“The market is moving a lot faster than supply chains that have historically supported a very physical business have been set up to do,” he added, predicting more mega-campuses, which will stress power grids and potentially lead to political tension.
LLMs rely on the quality of the data, chips and algorithms they are trained on, which occur within data centers globally, with some powered by cleaner energy than others depending on their location.
The demand for AI, coupled with limited land availability, has pushed U.S. data center prices up by 19%, according to real estate firm JLL. Nvidia CEO Jensen Huang estimates that $250 billion will be spent annually on data center equipment.
Venturo added that rivals like Amazon, Google and Microsoft face similar challenges.
Trademarked content poses huge risks
Following OpenAI’s Sora, Adobe is developing its text-to-video tool later this year. But Ashley Still, Adobe senior vice president, creative product group and digital media growth, didn’t share the release date.
“Video is in its very early stages,” she said. However, the first use cases for this video tool will focus on creating storyboards and shorter-form content tailored for social media platforms.
Adobe’s stock content business totals over 400 million licensed assets, including photos and illustrations, among others, for Firefly customers like marketing agencies, Still said.
“Firefly … hasn’t been trained on data like scraping the internet, [so] you can’t create trademarked content,” said Still. “You know you’re not going to get sued, and that’s a selling point.”
