Exploring the Dynamics of AI Chip Manufacturing: A New Era in the United States
The world of artificial intelligence (AI) is rapidly evolving, and at the heart of this transformation are AI chips, the essential components that power these advanced technologies. Most of these chips are produced in Taiwan by the renowned Taiwan Semiconductor Manufacturing Company (TSMC). However, a new wave of startups is emerging, aiming to reduce the cost of AI by collaborating with U.S. manufacturers. This shift is seeing AI chips being produced at fabrication facilities in New York and Arizona.
Competing with industry giant Nvidia is a formidable challenge, particularly in the realm of manufacturing. Nvidia, along with most of its competitors, does not produce its own chips. Instead, they rely on the world’s most advanced chip fabricator, TSMC. While Nvidia largely controls the distribution of the latest and most powerful computing machines, TSMC determines the quantity Nvidia can sell. The intricate relationship between Nvidia and TSMC captivates the industry.
Despite the dominance of Nvidia and the unparalleled manufacturing capabilities of TSMC, some startups believe they can carve out a niche by leveraging the fluctuating dynamics surrounding Taiwan and tapping into U.S. chip fabrication facilities.
Positron AI: A New Contender in AI Chip Design
Founded by Thomas Sohmers in 2023, Positron AI has developed a chip architecture optimized for transformer models, the same models on which OpenAI‘s GPT models are based. Sohmers claims that Positron’s architecture, with its faster access to more memory, can compete in terms of performance and price for AI inference, which involves the computation needed to generate an answer to a query after a model has been trained.
While Positron’s system has “woefully less FLOPS” than an Nvidia GPU, as Sohmers humorously noted, the architecture is designed to make up for this with efficiency, benefiting both Positron and its customers.
Smaller Fabs: A Strategic Advantage
Positron’s chips are manufactured in Chandler, Arizona, by Intel-owned Altera. Intel acquired Altera, a specialist in a specific type of programmable chip, in 2015. Some early Positron employees and advisors came from Altera, bringing with them valuable relationships and trust. This early partnership has provided Positron with some influence over Altera’s direction and a more cost-effective, flexible manufacturing partner.
The cost of AI is driven by the chip itself and the power required to operate it. To reduce costs, Positron is looking beyond TSMC, which currently holds significant bargaining power. “Fundamentally, Positron is trying to provide the best performance per dollar and performance per watt,” Sohmers explained.
In contrast to other industries, AI offers a unique proposition: U.S. production is often cheaper. “In most other industries, made in the USA actually means that it’s going to be more expensive. That’s not the case for semiconductors — at least for now,” Sohmers said.
Many fabrication facilities are eager to enter the AI market, but they lack the technical expertise, prestige, or track record, making it challenging to attract customers. Startups, which often lack the high order volumes that confer market power, are well-suited for these fabs, Sohmers noted. These less in-demand fabs offer more favorable terms, which Sohmers hopes will keep Positron competitive on price.
“If I have some optionality going with someone that is behind but has the ambition to get ahead, it’s always good from a customer or partner perspective,” he said, adding, “It gives both leverage.”
Utilizing U.S. fabs has kept Positron’s funding needs within reasonable limits and made scaling easier, Sohmers said.
Groq: Another Challenger in the AI Chip Arena
Positron is not alone in its quest to challenge Nvidia. Fellow competitor Groq partners with GlobalFoundries in upstate New York, aiming to make a similar impact in the AI computing market by offering competitive performance at a lower price.
Challenges and Skepticism
Despite the potential advantages, there are challenges. Some investors have expressed skepticism, Sohmers admitted. As an engineer, choosing not to work with the best fab in the world can feel unusual. “You have a lot more faith that TSMC is going to get to a good yield number on a new design pretty quickly and that they have a good level of consistency while, at other fabs, it can be kind of a dice roll,” he said.
With a global supply chain, no semiconductor is immune from geopolitical turmoil or the shifting winds of trade policy. The advantages of moving away from the ongoing tensions between Taiwan, China, and the U.S. serve as a counterbalance to any skepticism.
Positron is also working on sourcing more components and materials in North America, or at least outside China and Taiwan. Sourcing from Mexico, for example, offers greater safety from geopolitical turmoil. The simpler benefit is that shipping is faster, allowing for quicker prototyping.
It has taken time, but Sohmers noted that the industry is beginning to recognize the need for more players across the AI space. “People are finally getting uncomfortable with Nvidia having 90-plus percent market share,” he said.
Got a tip or an insight to share? Contact BI’s senior reporter Emma Cosgrove at ecosgrove@businessinsider.com or use the secure messaging app Signal: 443-333-9088.
Originally Written by: Emma Cosgrove
