Amazon Web Services (AWS) will now serve as Anthropic’s primary cloud and training partner for their advanced generative AI models, known as Claude.
These models require substantial computing power for training and deploying, making securing expensive AI chips an essential priority for startups in this space.
What this means for the future of AI
Anthropic, the San Francisco-based AI chatbot developer known for Claude AI chatbot development, has experienced “a year of breakout growth” after recruiting top executives from OpenAI in 2020. Their most recent contribution – structured as convertible notes that will be invested gradually over time – doubles Anthropic’s funding and brings their total backing up to $8 billion.
Anthropic has also chosen AWS as its cloud and training partner, agreeing to use Amazon’s custom Trainium and Inferentia chips for its most advanced models. These processors are specifically designed to improve training performance through optimized software optimization and hardware efficiency according to an Annapurna Labs blog post by Anthropic and Amazon’s Annapurna Labs division.
Amazon is taking strides to keep up with tech titans such as Microsoft and Google in terms of generative AI research. Amazon has made one of its largest investments ever into an emerging AI research lab; setting a landmark example as industry shifts toward using machine learning in real-world applications.
Microsoft’s OpenAI Deal
The deal marks another exciting development in an intense contest among technology giants over who will control the future of artificial intelligence (AI). It shows Microsoft’s renewed dedication to AI research and its determination not to fall behind rivals such as Google is having an effect in real time.
Amazon’s investment makes them a key partner of OpenAI, whose new AI model Claude 3.5 Sonnet has garnered widespread acclaim and attracted major enterprises. Through this partnership, AWS customers will also have early access to fine-tuning their models using data unique to themselves – something no other AI provider provides.
In 2022, Altman’s tenure at OpenAI abruptly came to a close when the nonprofit company abruptly let go of him over disagreements regarding its direction. OpenAI at that time was spending over $2 billion annually on computing power — more than any startup had in history — and had entered into a deal with Microsoft requiring all its chips for 10 years.
Google’s DeepMind Deal
Google and DeepMind’s partnership is strategic for them both; Google relies heavily on AI technology for its core search and advertising businesses and also holds a significant interest in autonomous vehicles, robots and smart devices – with futurist Ray Kurzweil being recruited to oversee its AI efforts.
Google was said to have outbid Facebook for DeepMind, founded by neuroscientist Demis Hassabis (an erstwhile child prodigy in chess) and Shane Legg. DeepMind is known for cutting-edge research; among its achievements were creating the world’s first computer to beat human Go players and WaveNet, an artificially intelligent text-to-speech system used by Google Assistant.
One of the key concerns over this agreement lies with how DeepMind plans to use Royal Free’s health data. Under UK law, companies processing personal data must have a legal justification for doing so – Royal Free and DeepMind assert their collaboration is solely for direct care purposes but this may not reflect contractually constrained usage plans for their data.
Amazon’s Strategy
Amazon Web Services will serve as Anthropic’s primary cloud and training partner and work together on optimizing AWS’ Trainium chips to increase training speed. A minority stake will remain held by Amazon. This agreement establishes Amazon as a primary training partner.
Amazon’s move signals its commitment to enterprise AI, building upon existing strengths in this space. Low prices and wide selection are hallmarks of success for e-commerce retailers like Amazon; now these same tactics seem to be leveraged against rivals like Microsoft and Alphabet’s Google in the expanding field of generative AI. As businesses transition from experimentation with AI towards deployment, high performance, security, and scalability options such as Anthropic’s solutions can meet that demand; already many startups, large enterprises, as well as European Parliament have adopted Anthropic’s Claude models for experimentation or deployment purposes.