Microsoft (MSFT) has emerged as Nvidia’s biggest trading partner amid the artificial intelligence frenzy. The software giant bought most of the chip giant’s advanced Hopper chips in 2024. While AI chips are in short supply, Microsoft allegedly bought twice as many of Nvidia’s (NVDA) chips as the second-largest customer.
Nvidia Customers
Immediate reports indicate that Microsoft purchased 485,000 high-performance computing chips to enhance its AI capabilities. The purchase comes from the software giant, which invested over $13 billion in OpenAI, the parent company behind ChatGPT. Consequently, Microsoft has emerged as Nvidia’s biggest customer, accounting for the most significant share of its chips revenue.
Microsoft is investing enormous financial and computational resources in creating the systems that it believes will be the defining technology of the future, as evidenced by its enormous stockpile of Nvidia chips. When OpenAI released ChatGPT-3.5 in November 2022, it ignited the current AI arms race. Since then, significant sums of money have been invested in R&D by Nvidia’s other important clients.
Tencent and TikTok owner ByteDance are Nvidia’s second-largest customers for advanced chips. While the two purchased 230,000 chips, Meta Platforms is the second largest US Company, and Nvidia generates most of its revenue. Amazon (AMZN) and Alphabet (GOOGL) are the other largest customers, with about 196,000 and 169,000 chips purchased.
While Hopper accounts for the most significant share of purchases, Nvidia is replacing it with the Blackwell chip. Nevertheless, it is still unclear whether the Blackwell chip will enjoy the level of success that Hopper has enjoyed over the past year. Meta Platform (META), Amazon, and Google are all developing their own chips to reduce their reliance on the expensive Nvidia Chips.
Salesforce AI Purchases
Meanwhile, Salesforce is in the process of hiring 2,000 people as it looks to accelerate the sale of its artificial intelligence software. It is double the number of people the company planned to hire last month, signaling its aggressive push to boost revenue within its artificial intelligence division. The hiring spree comes nearly two years after the company declared it was lying off over 7,000 workers to reflect the state of the economy better.
According to Salesforce, customers will be able to access the second generation of its Agentforce technology, which creates and runs AI agents, early next year. Using all available data, Agentforce will be able to respond to complex queries in Salesforce’s Slack communications app.
OpenAI vs. Google
Separately, OpenAI is getting ready to introduce a new artificial intelligence model that can reason more sophisticatedly than its current models, as it takes the fights to Alphabet Inc.’s Google. In order to solve more complicated multi-step problems, the new O3 model takes longer to calculate an answer before answering user inquiries. Additionally, the business will launch the o3-mini, a scaled-down version of the model.
OpenAI revealed some preliminary information about how its reasoning model, o3, outperforms o1, which it unveiled in September when responding to challenging questions about subjects like coding. As part of its procedure before releasing new software, OpenAI is also requesting applications from safety and security researchers to test the models. OpenAI Chief Executive Officer Sam Altman stated at the event that the company intends to release the o3-mini model at the end of January and the o3 model soon after.
