-
Concerns about hyperscalers gaining monopoly power over AI technology are growing
-
Cockroach Labs CEO Spencer Kimball thinks their power won't last
-
Open-source options could be AI's saving grace
After spending years simmering on the back burner while companies waited for economies of scale to emerge, artificial intelligence (AI) reached a boiling point over the past 24 months. But to ensure rapid AI adoption doesn’t exacerbate the monopolies that emerged in the cloud era, it will need open-source challengers, Cockroach Labs CEO Spencer Kimball told Silverlinings.
Many organizations let a craving for convenience drive their cloud adoption, Kimball said. The subsequent vendor lock-in has continued to trouble the industry throughout 2023.
The open-source database company’s recent report, “The State of the Multi-cloud,” found that vendor lock-in was the most cited reason for going multi-cloud among large companies. The report zeroed in on “the conservative outlook” amongst cloud architects and engineers “and it is very multi-cloud, not just in terms of what they want to do for the future, but what's already the reality,” Kimball said.
Lock-in struggles have only worsened as large cloud companies — Amazon, Meta, Microsoft and the like —have grown their controlling market share. Throw 2023’s poster child, AI, into the mix, and suddenly the Big Tech has the potential to become far more Orwellian.
Indeed, AI isn’t cheap. The lofty price tag for effective developments naturally lends cloud colossi the advantage. Yet these giants appear split in their perspective on the open-or-closed dilemma, reflected in lobbying efforts with regulators.
Taking sides
Meta and IBM are advocates for an “open-science” ecosystem, recently forming the “AI Alliance.” Rival players like Google, Microsoft and OpenAI offer a contesting closed perspective, forming a separate group called the “Frontier Model Forum.”
The idea of hyperscalers gatekeeping the technology has prompted industry voices to speak up with worry that AI will only worsen Big Tech’s power in 2024 and beyond. Meta’s chief AI scientist, Yann LeCun called out “massive corporate lobbying” from these companies that would set up regulatory guardrails in favor of their models, as reported by ABC News.
“Right now, there are only a handful of companies with the resources needed to create these large-scale AI models and deploy them at scale,” encrypted messaging app Signal President Meredith Whittaker told CNBC. “And we need to recognize that this is giving them inordinate power over our lives and institutions.”
Kimball partially agreed with that outlook. “It's clear that there's a very small AI club in terms of the people that can train the models that we're all aware of right now.” These multi-trillion-dollar companies are becoming indispensable, he added. “We're going to accelerate [their] dominance by giving them exclusive licenses to essentially create the foundations for putting intelligence into all products? This seems dystopian to me.”
Still, Kimball maintains a sense of “long-term optimism” around the open-source ecosystem being established in these early AI days.
Stepping up
“The idea that large language models are going to be confined to the multi-trillion-dollar companies is probably not accurate,” he argued.
“I think we're going to see a huge surge in open source and also in challengers” that will step up to the plate, Kimball added.
He praised a company he was unable to disclose for taking an “extremely novel” approach to how it was building its large language model. It has raised a staggering “$600 million in a seed, he said.
“It would be a mistake to think that there's a straight line sort of path to maintaining the dominance that these big players currently have… that's why I am encouraged to see that investors are willing to bet on challengers,” he said.
Kimball circled back to the lobbying hyperscalers, or the “small anti-AI, AI club,” saying that he hopes regulators and legislators “keep an open mind, which will be hard to do.” But he believes AI technologies need to be in the open-source world.
“I think it's a mistake to believe that they can be controlled… I've always been a lifelong proponent of open source,” he concluded. “I think that leads to better outcomes for humanity.”