By playing, you agree to uListen's Terms and Privacy Policy and YouTube's Terms of Service.
Masters of ScaleMicrosoft CEO Satya Nadella on AI, the metaverse & remote work | Masters of Scale Summit 2022
Summaries, chapters, transcripts, and FAQs on this page are made by uListen, not YouTube.
Great conversations run long. Their best ideas don't have to.
uListen is free on iPhone, iPad, Android, Chrome, and Claude.
See howCHAPTERS
Chapter summaries
Reid's Welcome and Why Microsoft–LinkedIn Worked
Reid Hoffman opens by saying that when he and Satya Nadella discussed combining Microsoft and LinkedIn, he knew the missions and cultures were aligned. What he hadn't known was that working with Satya would be part of the magic. Satya jokes about adding the praise to his board evaluation.
- •Reid saw the two companies' missions as friends, both about elevating human potential and productivity
- •Reid says he did not fully anticipate the value of working with Satya
- •Reid praises Satya for taking responsibility for shaping the world seriously
- •Satya jokes that Reid should put the praise into his board evaluation
Doing More with Less: Technology for Global Challenges
Reid asks how technology can help solve large-scale problems such as wars, pandemics, markets and climate change. Satya contrasts the Popular Electronics and Newsweek covers from Microsoft's founding month and says growth now has to be paired with protecting the planet and increasing trust and equity.
- •Popular Electronics and a Newsweek energy-crisis cover appeared the same month, a juxtaposition Satya says is back
- •Software is the most malleable resource for taming these constraints
- •Economic growth must not break the planet
- •Social cohesion and addressing inequities are needed too
- •Microsoft's mission: pick both magazine covers and go to work on them
Copilot and Amplifying Human Capability
Reid asks about Copilot and what it means for developers and all professionals. Satya describes GitHub Copilot as an AI pair programmer that makes him feel like a super programmer, notes that around 30-40% of code is AI-generated in the data they have, and says users stay in control. He also cites in/on/off-the-loop designs and a Copilot for chemistry in the energy transition.
- •Scaling compute is producing more intelligence in large multimodal models
- •GitHub Copilot generates code, explains code in English and can translate C++ to Rust
- •Around 30-40% of code is AI-generated in the data, with the human still in the loop
- •A Copilot for everything, such as Autodesk Maya in architecture and design
- •The energy transition requires compressing roughly 150 years of chemistry into a few decades
Adapting Skills and Careers: Excel-Class Skills for AI
Reid asks how people and organizations should adjust to these tools. Satya recalls first using a spreadsheet and says the modern equivalent is domain experts using low-code and no-code tools, assisted by AI, to build automation and apps.
- •Excel changed people's relationship with numbers
- •There will always be elite AI skills, but Excel-like skills spreading across every domain are most exciting
- •During the pandemic domain experts turned to robotic process automation and low-code tools
- •Power Platform can turn a napkin design into an app
- •This turns domain skill into digital skill
The Metaverse and Digital Twins
Reid raises the metaverse, starting with digital twins. Satya explains that software has always created digital twins by reference, whereas the metaverse digitizes places, things and people by value, enabling embodied presence without being there.
- •Bill Gates's idea that there is only one software category: information management
- •Digitizing by value rather than by reference
- •Embodied presence is the killer app
- •Pandemic-era digitization of warehouses and factories enables simulation, less waste and remote debugging
How the Cloud Powers the Metaverse and AI
Reid asks how the metaverse and cloud go together. Satya says the cloud has created an abundance of compute, and with Moore's law running out of gas new system architectures are needed, such as a new type of supercomputer for training large AI models.
- •Cloud is the first time you can throw lots more compute at any problem
- •Training large AI models required a new type of supercomputer
- •Different workloads like real-time streaming, visualization, training and inference need different architectures
- •Cloud means compute systems that change with the workload
The Flexibility vs. Connection Paradox in Hybrid Work
Reid asks how Satya leads Microsoft through the tension between wanting flexibility and wanting more human connection. Satya notes that many frontline workers never had remote options, then describes the productivity paradox and says leaders must use data, align goals and learn to create occasions.
- •About sixty percent of the workforce could not work remotely
- •Eighty-five percent of managers think employees are slacking while eighty-five percent of employees think they work too hard
- •Leaders should use data rather than dogma and bring clarity to outputs and measures
- •People come for people, not policy, so leaders must act as event managers
- •Re-recruit and re-energize people; a great 2019 leader can't just wish for 2019
Closing Advice for Leaders: Optimism and Energy
Reid asks for one message for leaders about a changing world. Satya says the world needs optimism and energy from leaders, and that the measure of a leader is bringing clarity, generating energy and solving over-constrained problems.
- •Leaders should offer optimism and energy
- •A leader brings clarity into a situation
- •A leader generates energy
- •A leader solves over-constrained problems
