On Nvidia GPUs: “It’s almost like talking about drugs: ‘I know a guy who has H100s.’”

The demand for graphics processing units (GPUs) in the tech industry, particularly for artificial intelligence (A.I.) applications, has led to a desperate search for these essential components. Start-ups and investors are going to great lengths to obtain GPUs due to their role in powering A.I. products that analyze vast amounts of data. The scarcity of GPUs began when the rise of A.I., fueled by chatbots like ChatGPT, created a surge in demand, causing a shortage in the market. Nvidia (NASDAQ: NVDA) further exacerbated the scarcity because it is the main supplier for these chips. Meanwhile, the company has scaled a $1 trillion valuation and is expected to report record financial results. (The New York Times)

Companies are struggling to access GPUs through cloud computing services due to long waitlists, which has led to technical tips being shared on social media and various creative strategies to secure Nvidia GPUs. Some venture capital firms are using their connections to purchase chips and offer them to portfolio companies. Start-ups are pooling resources to buy and share clusters of GPUs. Even highly technical figures like Eric Jonas, who left academia to start an A.I. company, are resorting to unconventional methods to acquire GPUs.

Despite challenges, some innovators are finding solutions. A project called the San Francisco Compute Group plans to let entrepreneurs and researchers buy access to GPUs in small amounts. This initiative is being supported by a specialized form of debt that uses GPUs as collateral.

Index Ventures has partnered with Oracle to provide Nvidia’s H100 and A100 chips to their portfolio companies for experimentation at no cost.

However, the GPU shortage has created a stark divide between well-funded companies with ample GPU access and those that are struggling to secure these vital components.

Related Story:  Chinese Internet Companies Place Billions Of Dollars’ Worth Of Orders For Nvidia AI Chips