Chart of the Week
The world is currently experiencing an arms race of artificial intelligence that has led to hyper realistic videos questioned on their authenticity and independent written pieces questioned on their “organic” intelligence way of structure and reasoning.
The top tech companies with a combined market cap over 15 trillion USD are spending significant amounts of time and money into research and development (R&D) for the fastest, smartest and best AI model for each task given to it.
Over the last decade over 700 models have been released that have significant computing power (greater than 10^23 FLOP), but releases peaked in 2021 at 105 units with these computing power released and then consistently released only 61 models every year from 2022 onwards. There are several reasons for the dip and plateau which include rising costs of development into breakthrough AI, resource constraints like GPU shortages and quality over quantity shift to focus on improving existing models rather than developing new ones. Make no mistake, there are still millions of public AI models uploaded annually, but they do not compete in computing power to these.
Google, Microsoft and Amazon all had similar R&D spending in 2015 (around 12 Billion USD), but have significantly diverged in amounts over the decade with Amazon leading the charge at 90 billion USD in R&D in 2024 focusing on AWS and Alexa improvements. Following behind Amazon is Google, investing 50 Billion USD in 2024 focusing on Google Search, Google Cloud and Gemini. Microsoft did not follow the intensive R&D spending in terms of amount, but rooted for an indirect path via a partnership with OpenAI. Microsoft still invests in AI through their Azure Cloud and Copilot tools. In 2024 they only spent 32 Billion USD on R&D.
Meta with the smallest market cap started with around 5 Billion USD into R&D, but surpassed Microsoft in 2020 and in 2024 invested around 50 Billion USD into R&D
Nvidia makes the GPUs required for AI development that all these companies use. Even though they are the largest in market cap, they only spent around 12 billion USD in R&D in 2024 that is used for GPU/chip development for AI acceleration.
Sam Altman recently said that AI is in a bubble and investors are overexcited about AI. With the top tech companies having between 60 to 90% of their capital expenditure in AI and the same amount into the R&D of their models, the bubble will continue to grow.
*Liaan Burger is a data and research analyst at Cirrus Capital.**
The top tech companies with a combined market cap over 15 trillion USD are spending significant amounts of time and money into research and development (R&D) for the fastest, smartest and best AI model for each task given to it.
Over the last decade over 700 models have been released that have significant computing power (greater than 10^23 FLOP), but releases peaked in 2021 at 105 units with these computing power released and then consistently released only 61 models every year from 2022 onwards. There are several reasons for the dip and plateau which include rising costs of development into breakthrough AI, resource constraints like GPU shortages and quality over quantity shift to focus on improving existing models rather than developing new ones. Make no mistake, there are still millions of public AI models uploaded annually, but they do not compete in computing power to these.
Google, Microsoft and Amazon all had similar R&D spending in 2015 (around 12 Billion USD), but have significantly diverged in amounts over the decade with Amazon leading the charge at 90 billion USD in R&D in 2024 focusing on AWS and Alexa improvements. Following behind Amazon is Google, investing 50 Billion USD in 2024 focusing on Google Search, Google Cloud and Gemini. Microsoft did not follow the intensive R&D spending in terms of amount, but rooted for an indirect path via a partnership with OpenAI. Microsoft still invests in AI through their Azure Cloud and Copilot tools. In 2024 they only spent 32 Billion USD on R&D.
Meta with the smallest market cap started with around 5 Billion USD into R&D, but surpassed Microsoft in 2020 and in 2024 invested around 50 Billion USD into R&D
Nvidia makes the GPUs required for AI development that all these companies use. Even though they are the largest in market cap, they only spent around 12 billion USD in R&D in 2024 that is used for GPU/chip development for AI acceleration.
Sam Altman recently said that AI is in a bubble and investors are overexcited about AI. With the top tech companies having between 60 to 90% of their capital expenditure in AI and the same amount into the R&D of their models, the bubble will continue to grow.
*Liaan Burger is a data and research analyst at Cirrus Capital.**
