Coatue: The Path to General-Purpose Robots

Coatue: The Path to General-Purpose Robots

Coatue Management, an investment firm, recently released a comprehensive report on the current state and future prospects of AI humanoids and robotics.

?Robotics has evolved from early industrial applications (with the first industrial robot used by GM in 1951) to becoming an integral part of our homes and workplaces. Despite impressive demonstrations, the reality of robotics has faced challenges in functionality, showcasing that robust performance takes numerous iterations to perfect. We are on the cusp of a new era, with AI-driven general-purpose robots demonstrating adaptive learning capabilities across diverse tasks and environments. This shift from single-purpose to multipurpose robotics signifies a monumental leap in technology, promising widespread implications across sectors.

?

Will We See a ChatGPT Moment for Robotics??

Coatue believes not and lists the reasons:

  • Physical limitations to adoption
  • High costs
  • Nascent ecosystem

?

I respectfully disagree with their conclusion. The advancement in AI will overcome limitations in dexterity and spatial awareness, allowing robots to handle failures effectively. AI's understanding of the world and its surroundings will enable robots to provide general-purpose functionality.

?In the next 10 years, we can expect the cost of all the elements required to have humanoid robots in homes to decrease by 75-80%, if not more. Compare the iPhone 1 to the iPhone 15. Although the iPhone 15 is more expensive than the iPhone 1, you get several orders of magnitude more value with the latest iPhone versus the first one. In a decade, every part, from chips to LiDAR to micro-engines and other components required for robots, will significantly increase in performance while decreasing in cost/performance ratio.

?Robots will be less dependent on their own ecosystems as they will thrive on the AI ecosystem. Recently, xAI started the Memphis Supercluster, the most powerful AI training cluster in the world with 100k liquid-cooled H100s on a single RDMA fabric. Tesla plans to have genuinely useful humanoid robots in low production for Tesla's internal use next year, with hopes for high production for other companies by 2026.

?

If you haven't seen this video, that I shared in one of my previous articles, that demonstrates improvements in robotics based on software updates only (adding ChatGPT), please do:

[https://x.com/Figure_robot/status/1767913661253984474]


Future: A Thriving Robotics Ecosystem?

The future foresees a thriving ecosystem for robotics, driven by significant advancements in AI, machine learning, and data analytics. Investment in AI robotics startups has seen a marked upsurge, signifying investor confidence in the sector's potential. Moreover, the democratization of robotics, led by advancements in AI, suggests the next wave of innovation could see robots becoming ubiquitous in everyday life.

?If you're still not convinced, ask yourself this: Do you have a Roomba? How many people do you know that have a Roomba or a similar type of robot? I have one. The problem with it is that it serves a single purpose. If the parts for a robot today cost $1 million, in 10 years, these will cost $100k. Do you think that's still too expensive? Do you use any help around the house, someone to mow your lawn, clean your house, or iron your shirts? How much do you pay for it? A thousand dollars a month? Two thousand? Now multiply this by 60 months for a 5-year payment program. Wouldn't you want to transfer that money and get yourself a robot that will not only mow the lawn, iron your shirt, or cook your food but can move all furniture from one room, repaint the room and bring back all the furniture, and fix your car while you're sleeping? Who wouldn't want that?

要查看或添加评论,请登录

Kresimir Furkovic的更多文章

社区洞察

其他会员也浏览了