AI summary by TwinMind

The Race For AI Robots Just Got Real (OpenAI, NVIDIA and more)

ColdFusion

21m 26s

Figure AI Company Background

  • Founded 2022, California-based tech startup focused on general purpose humanoid robots
  • Raised $70 million within first year, recently raised $675 million
  • Current valuation: $2.6 billion (unicorn status)
  • Investors: Microsoft, Nvidia, Jeff Bezos, Intel Capital
  • Mission: "expand human capabilities through advanced AI"
  • Founder: Brett Adcock (previously founded Vety - acquired for $100 million in 2018, Archer Aviation - publicly traded electric aircraft company)
  • Team includes veterans from Boston Dynamics, Google DeepMind, Tesla

Labor Market Problem Statement

  • 10 million unfilled jobs in United States
  • 7 million openings in essential roles (warehouse, transportation, retail)
  • Only 6 million people available to fill positions
  • Key warehouse suppliers predict running out of people to hire by 2024
  • Goal: humanoids join workforce with ability to think, learn, interact safely alongside humans

Figure-01 Robot Technical Specifications

  • Partnership with OpenAI announced February 2024
  • Integration with large multimodal models for natural language processing
  • All actions in demo were learned, not teleoperated
  • Equipment: cameras, microphones
  • Processing: transcribed text + images sent to OpenAI multimodal model
  • Mechanical action rate: 200 Hz (5 millisecond reaction time vs human 250 milliseconds, cat 30 milliseconds)
  • Uses combination of neural networks for audio/visual input processing

Demo Capabilities Shown

  • Natural conversation ability with speech hesitations (humanlike imperfections)
  • Object recognition and reasoning (gave apple because only edible item)
  • Task completion (putting dishes in drying rack)
  • Explanation of actions taken
  • Real-time, unaltered performance claimed by company

Technical Limitations

  • Latency between voice commands and responses (likely OpenAI server delays)
  • Instructions must go to server and return
  • Same delay issues observed in ChatGPT voice
  • Walking ability still limited
  • Concerns about emergency response scenarios (10-second delays problematic)

Competitive Landscape

Digit (Agility Robotics)

  • Backed by Amazon
  • Cost: $250,000 per unit
  • Can lift 35 lbs, human-sized reach
  • Designed for OSHA-compliant warehouse work
  • Amazon deploying in warehouses, aiming for 10,000 units/year production
  • Amazon investing $1 billion in industrial technologies

Aptronic

  • Designed for warehouse goods movement
  • 22 hours/day operation (battery swaps every few hours)
  • Deliveries start 2025
  • Working with NASA on Valkyrie robot for dangerous environments/space walks

Tesla Optimus

  • Demonstrated clothing folding (initially not autonomous)
  • Recent footage: carrying boxes, watering plants, walking, crouching, handling eggs
  • Benefits from Tesla self-driving knowledge
  • Lacks natural language aspect

Boston Dynamics Atlas

  • Most agile robot available
  • Focus on mobility, performs backflips
  • Less finger dexterity compared to Figure-01
  • Figure-01 has 5 fingers with 3 joints each for intricate tasks

Nvidia's Project Groot

  • General purpose foundational model for humanoid robots
  • Helps robots understand natural language, emulate human movements
  • Quick learning of coordination, dexterity, navigation skills
  • Training via simulation in Isaac Sim
  • Partnerships: Boston Dynamics, Agility Robotics, Aptronic, Figure AI
  • Industry standard forming: OpenAI (models) + Nvidia (hardware/software) + various robot manufacturers

Industry Transformation Context

  • COVID-19 triggered robotic sales surge, hitting record high 2022
  • Previous decades: robot instructions required hand-coding
  • Current: generative AI enables dynamic environment understanding
  • Automation progress reversed expectations - cognitive jobs impacted before manual labor

Employment Impact Statistics

  • 2017: 1 in 5 companies used AI (McKinsey)
  • Current: half of all companies use AI
  • Goldman Sachs 2023: 300+ million jobs could be lost/diminished by AI
  • Most at risk: legal, architecture, engineering, business/financial operations, management, sales, healthcare, art/design
  • Robot component prices dropping drastically

Commercial Deployment

  • Figure AI signed agreement with BMW manufacturing
  • Testing Figure-01 at BMW plant in Spartanburg, South Carolina
  • No performance reports available yet on specific tasks

Future Implications

  • Robots don't need breaks, sleep, holidays, don't strike or complain
  • Consumer robots still far from affordable local store availability
  • Figure-01 represents worst version from company (only improvements ahead)
  • Competition intensifying across multiple companies
  • Debate: complete human replacement vs. alleviating menial tasks
  • Military applications likely future consideration

TwinMind

Get summaries like this for your own meetings

TwinMind takes notes in the background during meetings, lectures, and calls, then writes the summary for you.

Free plan to get started

No bot joins your calls

Phone, desktop & Chrome apps

Your next meeting, summarized like this

TwinMind listens in the background and turns meetings, lectures, and calls into searchable notes you can chat with. Start free.

Available on phone, desktop & Chrome

Your next meeting, summarized like this

TwinMind listens in the background and turns meetings, lectures, and calls into searchable notes you can chat with. Start free.

Available on phone, desktop & Chrome

Your next meeting, summarized like this

TwinMind listens in the background and turns meetings, lectures, and calls into searchable notes you can chat with. Start free.

Available on phone, desktop & Chrome