Artificial intelligence (AI) technology is driving the growth in server infrastructure investments. According to Stocklytics.com, AI server spending will spike to $430 billion by 2033 from only $40 billion in 2024.
The site’s financial analyst, Edith Reads, said, “The rise in AI server spending highlights the budding demand for AI capabilities and the significant shift in how businesses and governments approach data processing and machine learning.”
AI projected server spending in the next nine years
By the end of 2024, AI server spending will increase by over $9 billion from 2023 to $40 billion. Moreover, the server market will likely see more profit rises, with companies expected to spend more than $50 billion in 2025 and close to surpassing $ 70 billion in 2026, with a projected server spending of $67 billion.
The AI server market’s revenue is expected to reach a wholesome $88 billion in 2027, a 31% increase from 2026 and a 120% increase from 2024. By 2030, the revenue will shoot up to $194 billion, representing another 384% rise in just six years. Additionally, in 2031, AI server spending will spike to $253 billion and cross the $330 billion mark by 2032. By 2033, the server spending will have escalated by over 30% from 2032 and by over 975% from 2024.
The rising demand for servers and its growing impacts
To start, Super Micro Computer, Inc. (SMCI) plans to acquire additional warehouses and ramp up its rack-scale production to 5,000 units by the end of fiscal year 2024 in response to the heightened demand for its servers. Additionally, manufacturing giant Foxconn has announced its goal to make AI servers its next “trillion Taiwanese-dollar revenue product.”
Besides, tech giant Dell is already selling its servers at near-zero margins to remain competitive in the evolving market, disappointing some of its investors.
Tech firms like Amazon Web Services and Meta are continuously fueling the demand gap and planning to expand their proprietary ASICs, while Chinese companies like Alibaba, Baidu, and Huawei are actively developing their ASIC AI solutions. By the end of the year, these companies will drive the bulk of ASIC servers to almost 30%.
Besides, Nvidia’s next-generation GPU manufacturing has called for ramped-up production for more advanced servers that can accommodate them. NVIDIA’s next-generation Blackwell, including GB200 and B100/B200, can run massive AI models at 25 times lower costs than previous Nvidia hardware and are expected to drive server demand all through 2025. Google has already ordered more than 400,000 Blackwell chips despite continuous developments in its own custom hardware, Tensor Processing Units (TPUs).
Leave a Comment