Big-Tech Silicon War Begins
All major tech and cloud providers are in a competitive race to develop custom, in-house silicon capabilities, primarily for handling machine learning and AI workloads. Recently, AWS unveiled its next generation of AI chips – Graviton4 and Trainium 2, positioning itself as a leader in this space.
The latest development places AWS in direct competition with other tech giants, each advancing their silicon solutions: Google with the TPU v5e, Microsoft with Azure Maia 100 and Cobalt 100, Apple with M3 chips, Meta with MTIA, and Tesla with its Dojo chips.
However, AWS claims to be far ahead of its competitors.?
AWS chief Adam Selipsky, at re:Invent 2023, said that the company has over 50K customers for Graviton, and its other cloud providers are still just talking about making them, but have yet to deliver their first server processors.?
This might be true in the case of Microsoft and Meta. At Ignite 2023 last month, Microsoft announced the launch of Azure Maia 100 AI Accelerator and Cobalt 100, its first in-house custom AI chips. The tech giant has been working closely with OpenAI on the design and testing phases of its AI chips. In May, Meta also unveiled its first version of a custom AI chip, Meta Training and Inference Accelerator (MTIA).?
When compared with Google, AWS is a generation behind. In August, Google announced the fifth generation of its custom TPUs, which is currently being used by thousands of startups, enterprises and research institutions. Some include Anthropic, Character.AI, Midjourney, AsemblyAI, Gridspace, and Lightricks, among others.?
What about Apple? Apple has a different approach to building custom, in-house AI chips. Apple Silicon houses a series of chips for almost all edge use cases. It focuses on enhancing users' experience in running machine learning and AI workloads, among other complex tasks. For instance, the recent release of M3 chips lets developers run large transformer models with billions of parameters on the MacBook, seamlessly.?
So far, it has not developed any custom AI chips for running its iCloud service and relies on hyperscalers like Google Cloud servers for storing data.?
Tesla, on the other hand, is looking to build a new home for Dojo chips to enhance the self-driving capabilities of its vehicles. Besides, it is likely to also use Dojo to sell cloud services to other companies in the coming days, similar to AWS, Google Cloud and Microsoft Azure. A few months ago, Tesla doubled its order for D1 chips with TSMC, as reported by the press in Taiwan.?
What will happen to NVIDIA? While a majority of tech companies continue to invest in building their in-house silicon capabilities, they are heavily dependent on NVIDIA to run most of their products and services. The in-house chips form only a small fraction of their operations, hence, the partnerships.?
However, competition cannot be ignored. Though tech companies are making quick progress on designing in-house chips, there is plenty of new business to go around. As NVIDIA continues to improve its existing chip capabilities (like in the case of L4 GPUs) and innovate, the requirement for collaboration only increases.?
NVIDIA will be more than just fine.
AI Forum for India. Our Discord Community for AI Ecosystem, In collaboration with NVIDIA.?Join Today >>
Best Firms for Data Scientists >>
Best Firms for Data Scientists is one of India's biggest workplace certification platforms in data science. To nominate your organisation, you may fill out the form here?.
People & Tech >>
领英推荐
Embracing Prudence and Pragmatism: Tredence’s Ethos for Innovation-led Product Development
When a majority of IT and data analytics companies were busy jumping on the generative AI bandwagon, Bengaluru-based Tredence focused on the due diligence side of things and their impact. In an exclusive interview with AIM, Pavan Kumar, senior manager at Tredence Studio, sheds light on the need for data product management, quality of data, the importance of user feedback and more. Check out the complete interview here.?
How eClerx is Shaping the Future by Transforming Business Processes with AI-ML
Mumbai-based IT consulting firm eClerx, in a recent interaction with AIM, dove deep into showcasing generative AI, metaverse, and NLP use cases, touching upon various aspects of the AI-driven transformation, alongside enabling conversational intelligence, market mix modelling for advanced marketing optimisation, effective budget planning and more. Learn more here.
Why Making Responsible AI Integral to GenAI Applications is Non-Negotiable
In an exclusive interaction with AIM, Krishnaram Kenthapadi, chief AI officer and chief scientist at Fiddler AI, shared the need for developing generative AI, responsibly. He also pointed out that it should not be an afterthought, alongside explaining the ethical pitfalls and reducing LLM hallucinations. Read the full story here.
AIM EVENTS >>
Rising 2024
AIM officially announces the sixth edition of Rising, a trailblazing conference focused on diversity and inclusion in technology, set to take place on April 4 and 5, 2024, in Bengaluru.?
The theme of this year’s conference will be ‘Beyond Binary: Redefining Diversity in Tech’. It is expected to witness over 1,000 in-person attendees and more than 50 influential speakers.
What are you waiting for? Register now!
AIM VIDEOS >>
re:Invent 2023?
[Reaction Video] In this latest video, AIM looks into some of the recent announcements made at re:Invent to understand their overall impact on the tech ecosystem.?
AIM NEWS >>