After crypto miners, now venture capitalists are coming for Nvidia GPUs

The graphics card market, which had somewhat stabilised in the last year after reeling from global supply shortages and hoarding by crypto miners, might again start heating up. And apart from gamers, this time, it has AI start-ups worried as well.
GPUs by Nvidia, one of the leading graphics card manufacturers, plays a crucial role in fueling today's thriving AI ecosystem. The A100 and H100 chips in its GPUs provide the necessary power for ChatGPT and other cutting-edge generative AI services to function.
The increasing number of AI startups and services has led to a high demand for these vital components, overwhelming Nvidia and its manufacturing partner TSMC, who are struggling to meet the supply. While Nvidia has experienced significant benefits from this, with a market capitalisation increase of nearly $200 billion in a single day last month, AI startups have faced challenges. Prices have skyrocketed, and shortages have become a common occurrence.
As a result, larger tech companies hold a significant advantage over smaller startups. To truly compete in the AI landscape, startups must train their own models using their own data, which often requires a substantial number of GPUs. Otherwise, they risk becoming mere applications on someone else's platform. 
Recognising this, Microsoft, along with its partner OpenAI, as well as Google and Adobe, is rushing to train massive foundation models using vast amounts of data, leveraging their billions of dollars in investments for this expensive undertaking.
Unfortunately, many startups are unable to afford such resources or secure the necessary chips. Even the tech giant Microsoft is facing hardware shortages, leading them to impose limitations on internal GPU access to conserve processing power for their AI-powered Bing chatbot and AI Microsoft Office tools. In light of these challenges, some venture capitalists are taking unconventional steps to provide assistance.
Nat Friedman, former CEO of GitHub, and Daniel Gross, a prominent investor in successful startups like GitHub and Uber, have made substantial purchases of thousands of GPUs to establish their own AI cloud service.
Named Andromeda Cluster, this system comprises 2,512 H100 GPUs and can train a 65 billion parameter AI model in approximately 10 days, as stated by the venture capitalists. While it may not be the largest model available, it is undoubtedly a significant achievement.
"Individual investors are doing more to support compute-intensive startups than most governments. (Very cool project!)," tweeted Jack Clark, a co-founder of AI startup Anthropic.
Friedman and Gross's actions demonstrate their commitment to assisting AI startups facing GPU shortages and resource limitations. By providing GPU resources through the Andromeda Cluster, they offer a lifeline to startups striving to compete in the AI landscape. This collaborative approach underscores the collective efforts being made to address the challenges encountered by AI startups in acquiring critical hardware resources.
Amid the increasing demand for Nvidia GPUs, innovative solutions like the Andromeda Cluster showcase the resilience and adaptability of the AI ecosystem. Industry players are coming together to foster a thriving environment for startups, ensuring that the necessary resources are available to drive continued growth and innovation.