Intel faces setback as cloud giant invests billions in rivals, while Arm startup emerges as a major player.
Larry Ellison, the founder and Chairman of Oracle, recently announced that the cloud computing giant is spending billions of dollars on GPU and computer chips from three main tech companies. Interestingly, Ellison made no mention of Intel, a significant blow to the emblematic x86 firm.
Ellison disclosed that Oracle will be buying GPUs from Nvidia, with a budget of billions of dollars. Additionally, the company plans to spend three times that amount on CPUs from Ampere and AMD. This news comes as Oracle aims to position itself as a major AI cloud computing player and grab market share from the dominant trio of AWS, Microsoft, and Google.
If Oracle’s GPU purchases from Nvidia amount to $2 billion, then the combined spending on GPUs from Ampere and AMD would reach $6 billion, bringing the total investment to approximately $8 billion. This figure is close to the $8.7 billion Oracle reported for its latest 12-month capital expenditure, almost doubling its spending from the previous year (ending May 2022) and more than quadrupling its FY 2021 CAPEX.
The announcement was made at an Ampere Computing event, where Oracle stated that it had invested more than $800 million in the startup. Ampere Computing specializes in designing Arm-based chips for the server market and was founded by Renée James, a former Intel President. Oracle further revealed that its database products will run on Ampere Computing chips, dealing a significant blow to the x86 architecture’s dominance.
Surprisingly, Intel was not mentioned during the event, making it the biggest loser in this scenario. As Intel has struggled to launch many-core processors, AMD has taken advantage of this weakness to secure another substantial win. Similar to Ampere Computing, AMD has focused on producing processors with a large number of physical cores. It recently unveiled its EPYC 9754 Bergamo CPU, boasting 128 cores and 256 threads. AMD’s 128-core/128-thread processor, the 9754S, is also targeted towards competing with Ampere Computing and Arm-based processors like AWS’s Graviton.
It is highly likely that Oracle has procured a significant amount of AMD’s new Instinct MI300 accelerator, which combines CPU, GPU, and high-bandwidth memory.
Additionally, Oracle’s multi-billion dollar deal with Nvidia will enable the company to obtain tens of thousands of specialized Tensor Core GPUs, such as the H100 Hopper or A100 models. These GPUs, priced at around $40,000 each (potentially less when bought in bulk), will empower Oracle to offer generative AI capabilities similar to Microsoft’s Azure and Google Cloud.
Oracle is also collaborating with Cohere, an AI platform specialist, to deliver generative AI using up to 16,000 Nvidia H100 GPUs per cluster. This partnership aims to accelerate the training of large language models while reducing costs.
The introduction of another major player in the cloud computing arena means increased competition, which in turn should lead to lower prices and foster innovation across various organizations, including startups and Fortune 100 companies.
Overall, Intel’s absence from Oracle’s recent investment plans and AMD’s advancements underscore the changing landscape of the tech industry. With cloud giants like Oracle expanding their partnerships with GPU and CPU manufacturers, it remains to be seen how Intel will respond and adapt to this growing competition.
