1.1 NVIDIA and Meta
It has been stated that NVIDIA and Meta are getting ready to expand their data center facilities in response to SK Hynix’s need for increased supplies of HBM and DDR5 memory from the company.
NVIDIA and Meta have plans to speed up the development of general artificial intelligence by rapidly expanding data centers; they will be demanding more DRAM supply from SK hynix
Companies like Meta have moved their attention toward the development of AI servers in order to benefit on the excitement that has been surrounding artificial intelligence (AI), as AI’s impact continues to grow. The launch of the AI Research SuperCluster, which included NVIDIA’s A100 GPUs and was announced by the firm exactly one year ago, is an excellent illustration of this concept.
In light of the fact that Meta intends to include GenAI into its widely used social media platform, the company is making headway toward the acquisition of cutting-edge technology in order to accomplish its objective.
SK Hynix site in Icheon
It has been reported by sources inside the industry that representatives from Meta made a trip to the SK Hynix site in Icheon. While there, the business examined its HBM and DDR5 memory division. In addition to that, Meta had performed a quality test for DDR5 memory modules as well, indicating that they were interested in the procedure. Meta is an established business partner of SK Hynix and focuses mostly on providing storage solutions for the latter’s server products.
However, as a result of the recent excitement around AI, Meta is currently focusing on acquiring DDR5 memory supplies. The corporation is, however, confronted with a significant obstacle, which might be described as a “unbalanced” supply chain.
It has been claimed that SK Hynix has expressed misgivings in response to Meta’s need for DDR5 memory, citing the fact that the business is now dealing with order backlogs as a result of the very high demand. This is rather obvious owing to the fact that every other firm competing in the GenAI race is working hard to acquire the technology required to run next-generation datacenters, one of the most important components of which is DDR5 memory. In other words, the competition is to see who can acquire the technology first.
In addition to Meta, it has been revealed that NVIDIA has plans to visit the facilities of SK Hynix in the near future, with the intention of having conversations over next-generation “HBM3e” memory for next AI GPUs.
The HBM sector is now seeing an economic resurgence, mostly owing to the relevance of HBMs in artificial intelligence graphics processing units (AI GPUs), in particular the NVIDIA H100s. NVIDIA had already taken a sample of the HBM3e memory, and based on what they saw, it seemed to be ready to be incorporated into next-generation AI products.
It is possible that suppliers may need to slow down their production in order to satisfy the enormous demand that would result from the significant production increase that Team Green intends to implement. NVIDIA is searching for a reliable supply chain in the hopes of alleviating the present order backlog crisis and meeting its goal of delivering 1.5 million to 2 million H100s in 2024. The company has this ambitious shipping goal.
The surge in artificial intelligence is a savior for AI-related suppliers like SK Hynix, which has been experiencing declining financial quarters. Since the beginning of the year, the Korean powerhouse has been experiencing economic difficulties, and the corporation has “bet” its future on the excitement around artificial intelligence. In light of the fact that it presently owns an 80% worldwide share of the HBM market, we have a good chance of seeing enormous revenue creation statistics in the next quarters.
[…] is a member of NVIDIA Metropolis and Inception. The third-busiest airport in India now uses AI’s vision AI […]
[…] They refer to it as the Multi-Modal Mobility Morphobot, or M4, and it was made possible by the NVIDIA Jetson platform for edge artificial intelligence and […]
[…] can configure NVIDIA Video Super Resolution on my MAG274UPF […]
[…] to a white paper co-published by SK Hynix and Intel, the DDR5 server DRAM used with Intel’s CPUs achieved performance levels that were […]