If you were Nvidia Corp. Chief Executive Officer Jensen Huang and you were laying out a few sector-shifting artificial-intelligence trends that could take shape over the coming months and years, you might include the following:
- Open-weight AI models, maybe from China, will continue to develop into a viable and cheaper alternative to closed and frontier models, thus becoming more popular for the majority of day-to-day tasks for typical businesses.
- Frontier AI companies will continue to make strides in diversifying the chips they use for inference, reducing their reliance on Nvidia.
- Prudent businesses will trend toward “sovereign” hosting and model building, rather than turning to the handful of powerful hyperscalers and frontier labs as they might do now.
For all these scenarios, Nvidia’s $13 billion Hugging Face acquisition, confirmed Thursday, represents a near-perfect hedge to how Nvidia’s business mostly operates today.
It places Nvidia at the center of the AI distribution ecosystem, helping it drive adoption of hundreds of thousands of diverse models, including its own. It would cushion the blow should the top AI labs, such as Anthropic or OpenAI, crash and burn — or simply capture less revenue than is expected today, slowing the demand for chips. It would help Nvidia avoid serious trouble should China win the AI race, or even just capture a large chunk of the business, with open-weight models. It builds on its strategy to support neocloud companies that rent access to AI computing power without going to one of the hyperscalers. Simply put, buying Hugging Face “allows Nvidia to have another competitive piece on the chess board,” tweeted venture capitalist Chamath Palihapitiya.
It strengthens Nvidia’s position of being model agnostic. Where the hyperscalers need businesses to use their AI, Nvidia’s motivation is to just get as many people as possible to use AI of any kind — knowing it sits behind most of the world’s infrastructure for doing so. Importantly, that includes CUDA, its programming model, that Huang has long maintained is a big part of his company’s competitive moat.
See more: NVIDIA Q2 Earnings & Outlook Matter More This Time. Here's Why
During the company’s most recent earnings call, held last week, Huang dedicated most of his remarks to talking up Nvidia’s strong position in open-weight models, stressing that he thinks prudent businesses will ultimately prefer to build their own AI, using open-weight models as the starting point. “Every major company and surely every country and every startup needs to build their domain specific, their proprietary AI, their proprietary alpha,” he said. “And the open models reaching frontier levels has made it possible, has enabled them to all do that.”
In that context, buying Hugging Face is highly complementary to Nvidia’s investments in neoclouds such as CoreWeave Inc. and Nebius Group NV. These now make sense beyond a circular financing vehicle to juice Nvidia chip sales. “We think more open-model workloads will seek private, dedicated and sovereign hosting outside hyperscaler bundles,” remarked Bloomberg Intelligence’s Vasu Kasibhotla, “which will reinforce neoclouds as AI’s neutral compute layer.”
The key term is “neutral” — the potential corrosion of which poses the biggest risk to the Hugging Face acquisition. The site, named after an emoji, became what it is today — a repository for more than 3 million open AI models — because of its independence from any of the main players, even as some of them became investors. As rumors of this week’s deal emerged, users raised concerns that Nvidia will use Hugging Face to funnel users aggressively to Nvidia models and infrastructure. Huang has insisted that won’t be the case. “Developers will choose the models they want, the frameworks they want, the clouds and inference service providers they want and the computing platforms they want,” he wrote in a blog post announcing the acquisition. “Nvidia compute will not be required to build on or deploy through Hugging Face.”
Parallels will inevitably be drawn, however, to Microsoft Corp.’s 2018 acquisition of coding repository GitHub Inc. Initially, developers praised Microsoft’s hands-off approach, but as the AI era beckoned, the company moved quickly to train its models on code hosted on GitHub, infuriating some developers. Later, it used GitHub as a marketing channel for its own AI coding tools.
Nvidia’s handling of Hugging Face will need to be delicate, lest it spoil the community that made the acquisition worthwhile. For now, investors clearly see the Hugging Face bet as a smart move. Nvidia shares were up 1.5% on Thursday on news of the deal’s confirmation. That’s not eye-catching until you consider that’s a more than $80 billion jump in market capitalization for a company that booked a $54 billion profit in the last quarter alone. That makes $13 billion small change for a big impact. The Hugging Face deal is a no-brainer and will likely be seen as a pivotal strategic move for years to come.
A message from Advisor Perspectives and VettaFi: Discover something new! Click here to register for our upcoming webcasts.
Bloomberg News provided this article. For more articles like this please visit
bloomberg.com.
More Insurance & Annuities Topics >
