In a rapidly evolving tech landscape, Google is taking significant strides to bolster its artificial intelligence (AI) infrastructure amid growing demand and ongoing chip shortages. On April 3, 2025, the tech giant unveiled its strategy to partner with emerging players in the market, particularly focusing on CoreWeave Inc., a company that has quickly gained traction in the AI sector.
The increasing appetite for computational power, driven by the surging development of generative AI models, has created a stark imbalance between supply and demand for graphics processing units (GPUs). The latest Blackwell chips from Nvidia, hailed as the pinnacle of AI technology, have become nearly impossible to obtain, with waiting lists stretching beyond 12 months. Google, like many of its tech counterparts, is grappling with these delays, which impact its AI-driven applications, including search, translation, cloud services, and digital assistance.
According to The Information, Google is currently in negotiations with CoreWeave to lease servers equipped with Nvidia Blackwell GPUs. This move represents a strategic pivot for Google, providing an immediate solution to the chip shortage directly from Nvidia. Additionally, Google is exploring the installation of its own Tensor Processing Unit (TPU) chips, developed in collaboration with Broadcom Inc., within CoreWeave’s data centers. This dual strategy not only aims to enhance available computing capacity but also seeks to diversify sources, reducing reliance on singular suppliers like Nvidia.
CoreWeave, a relatively new player in the market, has already made a name for itself, with its stock recently quoted at $40 per share, reflecting a remarkable 53% increase in value. Microsoft, a major client, accounts for 60% of CoreWeave’s annual revenue, which stands at approximately $1.9 billion. The company is also diversifying its client base, recently securing a five-year agreement with OpenAI valued at $12 billion, aimed at boosting the AI infrastructure that supports systems like ChatGPT.
This growing interest in CoreWeave underscores the explosive demand for AI capabilities, which is reshaping the technological landscape. Nvidia’s CEO, Jensen Huang, has predicted that the next generation of advanced generative AI models will require computing power at least 100 times greater than what current systems, such as ChatGPT, utilize. This forecast highlights the vast potential for growth within the sector and explains why companies like Google are pursuing aggressive strategies to secure their positions.
Despite the promising trajectory of the AI sector, both Google and Nvidia are navigating a challenging 2025, with their stocks experiencing declines of 17% and 20%, respectively, since the beginning of the year. These downturns are attributed to various factors, including high valuations and economic uncertainties influenced by trade tariff policies. Nevertheless, the excitement surrounding CoreWeave and its recent stock market rally indicates that investors still view companies linked to AI as viable long-term growth opportunities.
The broader context of Google’s interest in CoreWeave reflects a significant shift in the tech industry, where the race for AI infrastructure is becoming increasingly competitive. As companies strive to build scalable and reliable systems capable of meeting today’s demands and anticipating future needs, the hardware market for AI is poised for a redefinition. With computational power emerging as the new “digital gold” of the next decade, those who can provide it effectively are likely to lead the global digital transformation.
In a parallel development, researchers at Imperial College London have recently experienced a groundbreaking moment in scientific research, thanks to Google's AI tool, Co-Scientist. After spending a decade developing a hypothesis about antibiotic-resistant superbugs, the researchers were astonished when Co-Scientist, powered by Gemini 2.0, generated the same hypothesis in just 48 hours.
José R. Penadés, one of the researchers involved, expressed his amazement, even jokingly emailing Google to confirm that they hadn’t hacked into his computer. While the decade-long effort included extensive validation that the AI could not expedite, having the hypothesis generated so quickly could have saved years of research. Furthermore, Co-Scientist proposed four additional plausible hypotheses, including one the researchers had not previously considered, which they now plan to explore.
This instance showcases the potential of AI to revolutionize scientific research by analyzing vast amounts of data and identifying connections that might take human researchers significantly longer to uncover. Although there are concerns about AI “hallucinations” and the potential for job displacement, many, including Penadés, view these advancements as opportunities for scientists to focus on what truly matters in their work. He remarked, “This is definitely going to change science. I’m witnessing something spectacular, and I’m thrilled to be part of it.”
As Google continues to innovate and adapt in the face of challenges, its commitment to advancing AI technology remains clear. The partnerships with companies like CoreWeave and the groundbreaking applications of tools like Co-Scientist illustrate the potential for AI to not only enhance operational capabilities but also to transform entire fields of study. The future of AI is bright, and as the demand for computational power grows, so too does the promise of what can be achieved.
