One Of America's Biggest Entrepreneurs Emerges As A Key Nvidia Partner For Enterprise AI

Forbes

19-05-2025

Thai Lee speaks at the recent opening of the AI Labs in New Jersey.
Nvidia is opening another distribution strategy for its GPU capacity, by establishing a network of IT consulting firms as partners to test and sell AI solutions. The partners use new Spark machines and sell Nvidia infrastructure to enterprises, based on public information and the recent opening of one of the first partner sites.
One of the first partners to emerge is SHI, one of America's largest women-owned businesses. SHI, which has more than $14B in revenue, recently opened an AI and Cyber Labs in Piscataway, N.J., equipped with new Nvidia Spark machines. SHI also trained its staff to help companies build and test AI applications.
Nvidia has established a DGX SuperPOD Specialization Partner program, which suggests they're building a network of specialized partners who can deploy and manage their high-performance AI infrastructure, said David A. Bader, distinguished professor in Data Science at the New Jersey Institute of Technology, by email. ePlus, another IT services provider, has also become certified as a partner. The DGX SuperPOD is what Nvidia calls its AI supercomputer, accessed through the Cloud.
Nvidia is facing more competition from Chinese models for AI development, and needs to move quickly to cement its edge in enterprise AI.
'The work being done in SHI's AI & Cyber Labs showcases the accelerating shift toward AI-driven innovation across industries,' said Craig Weinstein, vice president of the America's Partner Organization, Nvidia, in an e-mailed statement. From AI agents helping people develop business strategies, to AI-assisted fraud detection, organizations are moving beyond experimentation to tangible results with SHI's labs, transforming how they operate and compete in the AI era.'
Working with trained partners – humans – makes sense since given the complexity of enterprise AI deployments. Many enterprise AI projects are stumbling on cost and data problems, according to outside researchers. The move also makes sense for Somerset-based SHI, which is expanding into AI-as-a-service. The new lab, opened to fanfare in April, will give teams from corporations access to the infrastructure needed to develop AI applications. They'll pay to rent time on Nvidia Sparks – new desktop AI development devices – at the Lab, along with receiving the advice of the SHI team based there.
'We believe we're developing capabilities that are going to be very useful,' said Thai Lee, CEO, in an interview. 'Many organizations that I talk to are afraid their business models today are not going to keep them alive.'
She said companies will need to adapt to the next wave of AI, including agentic AI and physical AI – meaning, the capacity of robots and autonomous vehicles.
Currently, companies that want to work on AI applications have to contract to engage with GPUs, which were housed in data centers requiring massive energy and cooling. Different models are now emerging for AI application development. The Spark, which is being sold as a personal AI computer, costs about $4,000, and contains Nvidia GPUs.
'These systems, powered by the Grace Blackwell platform, represent a democratization of AI computing power that was previously only available in data centers,' said Bader. 'For researchers, data scientists, and organizations exploring AI applications, having local access to this level of computing power through SHI's facility provides tremendous advantages. It enables experimentation with large language models and complex AI systems without the latency, cost, or data sovereignty concerns that can come with cloud-based approaches.'
The idea behind the Lab is that people will rent time there and with the assistance of more than 100 trained SHI staff, develop or test applications. Jack Hogan, vice president of advanced growth technologies at SHI, gave an example of how this has played out in practice.
'A large pharmaceutical company came to us, that is literally spending billions of dollars a year in AI infrastructure hosted in the could and on premises,' he said. 'They had a specific use case to ensure that the eight-figure purchase they were about to make was going to work.'
SHI charged them for a six-week engagement, with a price tag in 'the low six figures.'
'They ultimately proved to themselves that their data wasn't ready,' he said. 'They prevented themselves from making a substantial mistake in just going out and buying this equipment without knowing that it was going to work.'
SHI is also working with Nvidia on creating synthetic data for companies, so that even if they don't share their data, they can validate their use cases in the Lab.
Many enterprise-level AI projects are failing, according to research by Gartner. 'Cost is one of the greatest near-term threats to AI and generative AI (GenAI) success. Gartner estimates that more than half of organizations who are embarking on AI initiatives beyond 'everyday AI' are abandoning their efforts due to unexpected cost overruns,' according to a Gartner Research Note, titled Toolkit: Customer Calculation of AI and GenAI Multiyear Costs by Use Case, issued in March 2025.
About 30% of AI projects that make it past proof of concept will be abandoned due to poor data quality, inadequate risk controls, escalating costs or unclear business value, the research firm also predicted. However, at the same time, the successful AI projects will make a difference, it said reporting that through 2026, GenAI will reduce manually intensive data management costs up to 20% each year while enabling four times as many new use cases.
Nvidia, with a market capitalization of $2.6T, is one of the world's most valuable companies. By teaming up with it, SHI is further emerging as one of the most significant IT service companies in the United States,
Lee said SHI may open more labs across the country, working with Nvidia or others. 'The age of the GPU is here and it is going to completely transform the way that the computing world is been operating,' said Lee.
She also gave an interview on her leadership style, here.