While Nvidia’s chips currently rule the market, Meta is looking at alternatives.

The company has started testing an in-house chip for training AI systems, according to a report from Reuters. Meta’s chip is a dedicated accelerator, one source told Reuters, meaning it’s designed to handle only AI tasks and can save on power compared with conventional GPUs.

The move by Meta isn’t surprising: The chip is part of its Training and Inference Accelerator series, an effort to build a family of chips to power the company’s AI workloads. The company has also previously said that it intends to start using its own chips for training by 2026. The effort follows similar endeavors by fellow tech giants, like Amazon’s Trainium and Inferentia chips, and Google’s Tensor Processing Units.

As AI development threatens to devour a growing amount of power, the moves could be the vanguard of a search for energy-efficient means beyond conventional GPUs, said Justin St-Maurice, technical counselor at Info-Tech Research Group.

• With these chips, Meta has the opportunity to consider AI hardware from a “value per watt perspective,” which “doesn’t seem to be getting better with Nvidia,” said St-Maurice.
• “That brute force, high-energy use approach, we’re starting to see more and more examples of challenges to that,” he said.

“There’s a lot of work to be done around how energy intensive and how costly these models need to be to train and deploy,” said St-Maurice.

Meta’s move might also help get its massive infrastructure costs under control. The company expects to have $65 billion in capital expenditures in 2025 up from $38 billion in 2024, with much of that cash being directed toward AI infrastructure. Plus, the interest in in-house chips could signal a desire for self-reliance – and for more competition, said Rodolfo Rosini, CEO and co-founder of Vaire Computing.

“It’s an issue of injecting competition in one of the key areas,” said Rosini. “Because the moment that they exclusively buy from one vendor, that vendor can increase the prices.”

If other companies follow in the footsteps of the hyperscalers, it could spell trouble for Nvidia. While the chip giant currently dominates the market, Meta’s move indicates that it doesn’t want to “let Nvidia be the only innovator” and may be searching for efficient alternatives, said St-Maurice.

“The question is whether or not other companies will find other ways to do the same thing, more efficiently, more cost effectively, and start to create their own chips that can effectively do that without having to buy the premium Nvidia product,” said St-Maurice.

The problem, however, is resources. Most firms don’t have access to the same cash or capabilities as tech giants like Meta, Amazon and Google to design and build their own chips. Without the proper resources, Rosini said, high quality and performance are difficult to achieve. “To match the performance of Nvidia, you have to put in Nvidia money,” said Rosini.

ServiceNow announced an agreement this week to acquire enterprise AI startup Moveworks as the company expands its agentic footprint.

ServiceNow will buy MoveWorks in a cash-and-stock deal worth $2.85 billion, marking its biggest acquisition to date. Along with allowing ServiceNow to roll Moveworks’ front-end AI tools into its own offerings, the deal adds the startup’s team of 500 to the company.

“We have been focused on data, AI and workflow, and this (acquisition) accelerates a lot of the things we’re doing already,” Amit Zavery, president, chief product officer and COO of ServiceNow told CIO Upside.

Movework’s front-end technology suite provides users with an intuitive unified user interface that works across all parts of a business, said Zavery. One of its key offerings is its robust enterprise search tool, which creates a centralized way to quickly find relevant information from across a business’s systems.

• Good enterprise search tools can be hard to come by, said Zavery. While many enterprises want the ability to get a good sense of their internal data through enterprise search, these tools often face security and permissioning issues.
• Contextual understanding of the individual user is why Movework’s enterprise search product was so attractive to ServiceNow, he said. “What Moveworks has done is (create) a very well-thought out enterprise search which is specific to individual user requirements,” said Zavery.

The capabilities go hand in hand with ServiceNow’s ability to contextually automate tasks, he said.

“(ServiceNow) is the control tower, or the integrated glue for many companies today,” said Zavery. “Now, when we combine that with (Moveworks’) enterprise search and integrated user experience, we become much more valuable to customers.”

The purchase will also allow ServiceNow to expand its capabilities in customer relationship management, particularly customer service, which represents one of the “biggest and fastest-growing businesses for ServiceNow,” said Zavery. While Moveworks provides the “unified interface” for customers, ServiceNow’s tech helps them “through the journey of resolving issues,” he said.

The deal marks ServiceNow’s second acquisition this year: In January, the company announced a deal to buy Cuein, a conversational data firm, with a similar goal of accelerating its agentic AI plans. Rather than building competing tools, acquisitions like these help the company target the right markets at the right time, Zavery said. “We’ve been building a lot of things … We can’t build everything.”

Plus, the addition of Movework’s staff to ServiceNow is particularly advantageous given that AI talent is “at a premium today,” said Zavery. “We’re getting an opportunity to really bring in a very, very talented team who has been in the middle of AI evolution.”

As AI’s appetite for energy increases, some tech firms are scrambling to devise solutions.

IBM is seeking to patent a workload management system for “sustainable computing on (an) unstable power environment.” As the title of the filing suggests, the tech monitors tasks and power at a site, such as a data center, and adjusts workloads based on availability of “unstable” energy sources like wind power.

The system would collect information about the tasks at hand, such as the resource requirements and duration, and schedule them with the goal of maximizing usage of power from the unstable source.

Microsoft, meanwhile, is looking into energy monitoring on an individual level. The company filed a patent application for a system to monitor the power usage of different applications, providing a visual indicator for each one.

“There is presently no way for a user to make an informed choice between the applications based on environmental impact,” Microsoft said in the filing.

Patents like this aren’t new for Microsoft or IBM, and mirror those from other major tech firms: Nvidia, Intel and Google have all filed patent applications to manage data center energy usage.

The problem is that, as AI development demands an increasing amount of energy, the tech is on track to jack up power demand by 165% by the end of the decade. Some of these tech firms have already seen the impact: Google’s 2024 climate report noted a 48% rise in greenhouse gas emissions since 2019, and Microsoft’s sustainability report noted a 29% rise since 2020, both at least partially attributed to AI-related data center energy demand.

The question remains whether the ideas detailed in these patents would be enough to keep up with the rapid pace of AI development – and limit the carbon emissions it’s causing.