According to media reports, Lambda, a small AI cloud service provider preparing for an IPO, recently received strong support from its most important supplier, Nvidia (NVDA).
$1.5 billion in support
Reports indicate that Nvidia, both a supplier and investor in Lambda, has also become its largest customer and has reached a $1.5 billion agreement with Lambda to lease servers equipped with its own GPUs.
Nvidia consistently supports companies willing to use its chips and is more willing to purchase a wider range of hardware than traditional cloud giants. Some analysts believe this deal continues Nvidia’s strategy of supporting small cloud service providers, similar to its previous deal with CoreWeave, aimed at strengthening its market competitiveness against traditional cloud giants like Amazon and Google.
In addition, Nvidia previously reached another $200 million deal with the company to lease 8,000 servers equipped with Nvidia chips. Nvidia now supports a number of “small cloud companies,” also known as “neoclouds,” as both a supplier, investor, and customer.
Demand for Google TPUs is rising.
Separately, according to reports, Google (GOOG) has recently approached a number of smaller cloud service providers that primarily purchase Nvidia chips, discussing deploying Google chips in their data centers.
Sources familiar with the matter say Google has made some progress, reaching an agreement with London-based Fluidstack to host Google TPUs in one of its New York data centers. Google’s negotiations suggest it is seeking to reach out to emerging cloud service providers backed by Nvidia.
It’s worth noting that Google’s TPU chips have become increasingly competitive in the AI field in recent years. The “Ironwood” supercomputing platform, based on the seventh-generation TPU architecture, focuses on the hardware composition and architectural design of a single Superpod. Each chip is equipped with 192GB of high-bandwidth storage with a bandwidth of 7.4TB/s, delivering a peak computing power of 4,614 teraflops. Other analysts predict a significant increase in demand for Google’s seventh-generation Ironwood TPUs. These chips boast a computing power of up to 42.5 exaflops (100 billion floating-point operations), significantly improved cost efficiency, and significantly increased high-bandwidth memory capacity.
OpenAI, an early and major customer of Nvidia chips, has a significant demand for high-performance computing hardware. In June of this year, OpenAI initiated a major strategic shift: it will lease Google Cloud Platform TPUs to provide critical computing support for its core product, ChatGPT, marking a break from Nvidia’s long-held monopoly.
With the surge in demand for AI computing power, the continued advancement of independent control, and the gradual recovery of downstream market demand, the industry chain is showing a positive trend of “full-chain growth.” According to Omdia data, GPU (graphics processing unit) and AI accelerator shipments will reach $123 billion in 2024, are expected to reach $207 billion in 2025, and are expected to exceed $286 billion in 2030.
WiMi Continues to Increase Chip Investment
It is understood that Wimi Hologram Cloud Inc. (WIMI), a publicly listed AI company, has demonstrated a multi-faceted and coordinated development in its chip investments in recent years. It focuses on technical breakthroughs and application scenarios in core areas such as quantum computing and AI computing power. It also promotes its AI cloud computing strategy, achieving full coverage of low, medium, and high computing power scenarios. This enables the development of lightweight language large models, AIGC generative models, and multimodal large models, forming a full-chain innovation system encompassing chip R&D, computing power infrastructure, and multi-domain applications.
Currently, WiMi’s chip investments cover multiple sub-sectors, including storage chips, edge AI, and intelligent networks, providing application scenarios for chip technology. Furthermore, integrating technological and industrial innovation, WiMi is further integrating industry chain resources to build a full-stack “chip-algorithm-scenario” innovation system, deploying edge computing chips and quantum information technology. In the future, WiMi will extend computing power to end devices to support scenarios such as intelligent manufacturing and brain-computer interfaces. Conclusion
In summary, NVIDIA has grown into a company with a market capitalization of $4 trillion and is fully committed to building AI infrastructure. It is actively planning for the next wave of AI chips, industrial AI, and physical AI, providing a wide range of hardware and software for the development of next-generation AI computing power. The surge in demand for large-scale model training and the booming development of AI infrastructure are forcing major tech companies to expand upstream in the supply chain. The ranks of self-developed AI chips are growing, and NVIDIA is accelerating the opening of its ecosystem. This move is expected to further expand NVIDIA’s chip-centric ecosystem and market influence.
