AI summary by TwinMind

Why Apple is Quietly Buying AI Companies

ColdFusion

13m 31s

Apple's Silent AI Acquisition Strategy

  • Apple has bought 21 AI startups since 2017 (vs Microsoft 12, Meta 11, Google 8)
  • Spent ~$1 billion on AI ventures while remaining publicly silent
  • Acquisitions span self-driving tech, voice design, music generation, image recognition
  • Follows Apple's typical pattern: move last but decisively, revolutionize existing categories

Apple's Internal AI Development

  • Tim Cook revealed August 2023: Apple researching generative AI for years
  • Since 2020: assembled foundational models team (16 people, including former Google engineers)
  • Team head: John Giannandrea (former Google machine intelligence research head)
  • Created large language model "Ajax GPT" - outperforms GPT 3.5
  • Working on AI chatbot "Apple GPT" based on Ajax
  • Rumored mid-2024 release date, currently tightly controlled within company
  • Released frameworks/model libraries for Apple chipsets
  • Multiple teams working on various AI projects across company

Siri's Internal Dysfunction (2018-2023)

  • Past 5 years: Siri team devolved into "dumpster fire"
  • Engineers vs management bitter arguments about direction
  • No meaningful progress made

Key Problems:

  • Engineers denied tools to see user data/usage statistics
  • Leadership refused requests for more data to improve Siri
  • Apple executives insisted on on-device processing only (no server connections)
  • Disadvantaged Apple vs competitors using massive cloud AI models
  • Tim Cook + senior executives refused AI-generated responses
  • Opted for team of 20 writers creating pre-written canned responses
  • 2019: Design team rejected user feedback feature for Siri answers
  • Upper management wanted Siri to appear "all-knowing"
  • Engineers working blind, couldn't improve system

Staff Exodus:

  • Frustrated employees left for Google to work on large language models
  • Tim Cook personally pleaded for engineers to stay
  • Project "Blackbird" - attempt to rebuild Siri from ground up
    • Aimed for faster, lighter, more modular Siri
    • Axed by two senior Siri leaders who championed "Siri X" project
    • Siri X only focused on moving processing on-device
  • Vision Pro team so unimpressed with Siri they considered creating own voice system

Future Siri Improvements

  • iOS 18 expected to introduce significant AI updates
  • Apple's generative AI technology to significantly improve Siri
  • Large language models to make Siri smarter
  • Goal: Siri performs complex tasks from simple voice commands
  • Would threaten Rabbit R1's market position

Strategic AI Acquisitions for Siri:

  • Inductive: improve Siri's data processing
  • Voysis: help Siri understand natural language
  • PullString: make Siri features easier for iOS developers

Apple's Competitive Advantages

  • Over 1 billion devices already in user hands
  • Can leverage on-device/Edge AI (no cloud/internet required)
  • Faster, more secure, reliable performance
  • $200 million acquisition of Xnor AI for on-device processing
  • ARM-based chips with custom neural processing units
  • Complete hardware/software control enables tighter integration
  • Aligns with Alan Kay's philosophy: "People serious about software should make their own hardware"

Technical Breakthrough

  • Apple researchers squeezed large language model into MacBook
  • Runs 4-25x faster than standard models (GPU vs CPU focused)
  • Techniques include:
    • Storing model in flash memory instead of RAM
    • Can run models 2x larger than computer's physical memory
  • Research paper released detailing methodology

Market Analysis

  • Analyst perspective: Siri stagnant for years, breakthrough needed
  • OpenAI/ChatGPT not point of differentiation long-term
  • Apple's unique advantage: billion+ consumer devices
  • Features embedded in iOS push to hundreds of millions users overnight
  • Market access advantage no competitor has

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