Due to the increase in AI technology, there have been delays in the supply of AI-capable chips because demand has outpaced supply. To lessen their reliance on the main suppliers of GPUs, NVIDIA and AMD, worldwide companies like Microsoft, Google, and AWS are launching custom silicone production.
In consequence, APAC businesses may quickly find themselves utilizing a variety of chip forms in cloud data centers. The chips they choose will be influenced by the value and vendor relationships between the cloud vendors and the determine power and speed needed for various application workloads.
Big cloud manufacturers are making investments in practice silicon chips.
Compute-intensive things like training an Artificial huge language model demand a lot of computing power. Super advanced transistor cards from companies like NVIDIA and AMD have become very expensive and difficult to obtain as need for AI technology has increased.
The main hyperscale sky manufacturers have reacted by accelerating the development of custom silicone cards in 2023 and 2024. The initiatives will lessen their reliance on elitist vendors, enabling them to provide AI determine services to customers both nationally and in APAC.
With the launch of the Axion computer during its Cloud Next meeting in April 2024, Google made its first ever practice ARM-based Computers. The move up to producing its own CPUs, which builds on custom silicon work over the past ten years, is intended to support a range of common purpose computing, including CPU-based AI training.
The device is anticipated to enhance Google’s AI abilities within its data centre footprints and will be accessible to Google Cloud customers afterward in 2024 for Google’s fog customers in APAC.
Microsoft
Microsoft, also, has unveiled its unique first in- home tailor-made accelerator optimised for AI and conceptual AI tasks, which it has badged the Azure Maia 100 AI Accelerator. This is followed by the Cobalt 100, its unique ARM-based CPU, which was officially announced at Microsoft Ignite in November 2023. The company’s custom golden for AI has already been used for things like running OpenAI’s big language model, ChatGPT 3. 5. The global tech tycoon stated that starting in 2024, it would anticipate a more extensive deployment of Azure cloud data centers for consumers.
AWS
AWS invested in custom silicone bits in 2009 with a history. The company has today released four decades of Graviton CPU chips, which have been distributed to data centers all over the world, including in APAC. The processors were designed to increase the cost efficiency for sky workloads. Two years of Trainium and two decades of Inferentia for heavy learning and AI inferencing have been added to these for teaching 100B+ feature AI models.
AWS discusses the golden alternative for APAC fog customers.
At a recent AWS Summit held in Australia, Dave Brown, vice president of AWS Compute &, Networking Services, told TechRepublic the sky company’s cause for designing practice golden was on providing customers selection and improving “price performance” of attainable compute.
” Providing choice has been very important”, Brown said. ” Our customers can find the processors and accelerators that work best for their workload,” said the company. And with us producing our own custom silicon, we can give them more compute at a lower price”, he added.
NVIDIA, AMD and Intel among AWS chip suppliers
AWS has established long-standing relationships with major semiconductor chip manufacturers. For example, AWS ‘ relationship with NVIDIA, the now- dominant player in AI, dates back 13 years, while Intel, which has released Gaudi accelerators for AI, has been a supplier of semiconductors since the cloud provider’s beginnings. Since 2018, AWS has been placing chips from AMD in data centers.
Custom silicon is in demand as a result of rising costs
Customers moving to AWS Graviton in every single region, including APAC, has been a result of the cost optimisation fever that has gripped businesses over the past two years as the global economy has slowed, according to Brown. He said the chips have been widely adopted by the market — by more than 50, 000 customers globally — including all the hyperscaler’s top 100 customers. According to him,” The largest institutions are moving to Graviton because of performance benefits and cost savings.”
SEE: Cloud cost optimisation tools not enough to reign in cloud spending.
South Korean, Australian companies among users
Customers in APAC are using these options due to the widespread deployment of custom AWS silicon.
- Leonardo. Ai: The hyper- growth Australia- based image- generator startup Leonardo. Inferentia and Trainium chips have been used by AI in the training and inference of generative AI models. Brown said they had seen a 60 % reduction in inferencing costs and a 55 % latency improvement.
- Kakaopay Securities: South Korean financial institution Kakaopay Securities has been “using Graviton in a big way,” according to Brown. This has seen the banking player achieve a 20 % reduction in operational costs and a 30 % improvement in performance, Brown said.
Advantages of custom silicon for customers in the enterprise cloud
A growing range of compute options for enterprise customers in APAC could be beneficial, whether it is based on performance, cost, or suitability for various cloud workloads. Organizations could also be able to achieve sustainability goals with customized silicon options.
Improved performance and latency outcomes
The competition provided by cloud providers, in tandem with chip suppliers, could drive advances in chip performance, whether that is in the high- performance computing category for AI model training, or innovation for inferencing, where latency is a big consideration.
Potential for further cloud cost optimisation
Enterprises have faced a significant challenge with cloud cost optimization because growing cloud workloads have resulted in higher costs for customers. Customers have more choice when it comes to reducing overall cloud costs because more hardware options give them more choice when it comes to choosing the best compute.
ability to balance compute with application workloads
Enterprises will be able to better match their application workloads to the specific characteristics of the underlying hardware thanks to a growing selection of custom silicon chips, making sure they use the best silicon for the use cases they are pursuing.
reduced power, improved sustainability, and
By 2028, customers are expected to choose sustainability as their top priority when choosing cloud vendors. Vendors are responding: for instance, AWS said carbon emissions can be slashed using Graviton4 chips, which are 60 % more efficient. Custom silicon will help improve overall cloud sustainability.
