CoreWeave, a prominent AI compute provider, has injected significant optimism into the AI infrastructure sector, particularly benefiting Nvidia and other data center-related stocks. The company's recent insights on the extended lifespan of Nvidia's AI chips, shared during its earnings call and in a subsequent CNBC interview, bolster the investment case for artificial intelligence hardware.
This development provides greater assurance to data center operators like Amazon and Microsoft regarding the return on their current capital expenditures, potentially extending the ongoing capital expenditure cycle. Such sustainability is a positive indicator for future revenues across the entire supply chain, from chip manufacturers to power and electrical infrastructure providers. Furthermore, CoreWeave's commentary validates Nvidia's ambitious $500 billion financing initiative, announced in partnership with major financial firms.
The market reacted favorably to these developments, with Nvidia shares climbing 3%, networking provider Corning up 5.2%, GE Vernova rising 2.7%, and memory supplier Micron jumping 4.9% on Wednesday. CoreWeave itself saw its stock surge nearly 20% following what analysts described as a "true breakout quarter," with revenues exceeding expectations, losses narrowing, and an improved top-line outlook.
A significant concern for investors in the AI trade has been the potential for technological obsolescence, questioning the useful life of AI chips and thus the return on investment (ROI). However, CoreWeave's CEO, Mike Intrator, stated, "Older generations of GPUs are going to have a longer useful life than anyone anticipated." He added that these chips are being contracted for longer terms and at higher prices.
This assertion is supported by CoreWeave's finance chief, Nitin Agrawal, who revealed the company signed an A100 chip contract extending to 2029, despite the chip's introduction in 2020. The A100, part of Nvidia's Ampere generation, was instrumental in training early AI models like OpenAI's ChatGPT. Even with newer Nvidia generations like Hopper, Blackwell, and Rubin available, the older A100s remain in high demand.
Amazon CEO Andy Jassy also provided supporting data, noting that AI computing hardware, including servers and networking equipment, typically breaks even within three years and has a useful life of five to six years. Contracts are often secured for at least five-year terms. Any additional economic life beyond the initial contract term represents an embedded "call option" for investors, a scenario CoreWeave is experiencing with its A100s.
Intrator further elaborated on the factors enabling extended chip lifespans: Nvidia's hardware, its CUDA developer software, and CoreWeave's cloud delivery platform. While he emphasized CoreWeave's unique delivery solution, the foundational importance of Nvidia's hardware and software is key for investors. This suggests that hyperscale cloud providers like Amazon, Microsoft, and Google, as well as other major AI spenders like Meta Platforms, can also realize substantial ROI from older-generation chips.
The extended ROI potential translates to greater sustainability in capital expenditures for these tech giants, offering a margin of safety and more time to achieve profitable returns. This positive outlook benefits a wide range of companies, including semiconductor manufacturers (Nvidia, Broadcom, Intel, Micron), materials suppliers (Qnity), and infrastructure providers (Corning, Eaton, GE Vernova).
CoreWeave projects reaching at least 8 gigawatts of capacity by 2030, anticipating that demand will continue to outstrip supply for years to come. The extended useful life of AI chips is crucial for financing the massive buildout of data centers, a concept underpinning Nvidia's partnership with firms like BlackRock and Goldman Sachs. This initiative aims to create asset-backed securities using data centers as collateral, similar to mortgage-backed securities, providing a new avenue for funding AI infrastructure.
In essence, CoreWeave's positive outlook on chip longevity, combined with insights from Amazon, reinforces the view of compute as a viable and growing asset class. This perspective suggests the AI trade has considerable room for growth well into the future.
