Concepts framing why organizations move to cloud — innovation waves and the "burning platform" — plus the evolution of computing power (CPU → GPU → quantum) and the AWS offerings at each level, including a tour of Amazon Braket.

Innovation waves (Kondratiev / K-waves)

  • K-waves are hypothesized cycle-like phenomena in the global economy, closely connected with technology life cycles; each wave irreversibly changes society on a global scale 1:14:12
  • Historical waves: steam engine/cotton → railway/steel → electrical engineering/chemistry → petrochemicals/automobiles → information technology — cloud technology is the latest wave (with Web3/ML/AI possibly part of it or separate waves) 1:14:39
  • Each wave follows a supply/demand pattern of prosperity, recession, depression, improvement — the pattern tells you that you're in a wave and where you are in it 1:14:39 1:15:09

Burning platform

  • A burning platform is when a company abandons old technology for new technology with uncertainty of success, motivated by fear that the organization's survival hinges on digital transformation 1:15:27
  • Examples: dropping on-prem for cloud, or moving from cloud to Web3 1:15:27
  • aws.amazon.com → "Products" dropdown lists all offerings; each service page has details and a Pricing tab — checking pricing is something you'll do constantly (looking at pricing pages directly is easier than the calculator) 1:15:55 1:16:25 1:16:52
  • Documentation location changes as AWS updates its UI — searching "EC2 documentation" is often faster; in the console, just type the service name in the search bar to jump to it 1:17:17 1:17:48

Evolution of computing power

  • Computing power = the throughput at which a computer completes computational tasks 1:18:15
  • General computing: e.g., Xeon CPU processors — what you find in AWS data centers 1:18:15
  • GPU/tensor computing: ~50x faster than traditional CPU (metric borrowed from Google's tensor computing); used for specialized ML/AI tasks, not regular web workloads 1:18:40 1:19:09
  • Quantum computing: e.g., the Rigetti 16Q Aspen-4; ~100 million times faster, super cutting edge, few practical applications yet 1:19:09 1:19:38
  • New levels of compute add to — not replace — general computing 1:19:09

AWS offering per compute level

  • General computing → EC2 (Elastic Compute Cloud) with varied instance types/hardware 1:20:03
  • GPU/accelerated computing → AWS Inferentia chip, designed as a direct competitor to GCP's TPU; advantage: works with any ML framework, not just TensorFlow 1:20:03 1:20:29
  • Quantum computing → Amazon Braket (quantum computing as a service), delivered via a partnership with Caltech rather than AWS-built hardware 1:20:29 1:21:00

Amazon Braket walkthrough

  • Search "Braket" or "quantum computing" in the console; you may need to accept terms first 1:21:27
  • Offers multiple quantum devices with varying qubit counts, an algorithm library, ready-to-go notebooks connected to those devices, and "quantum tasks" 1:21:27 1:21:57
  • Pricing is per shot (each run) plus per task, varying by qubit family/device 1:22:26 1:22:44
  • Practical applications are extremely limited today — the point is that quantum research access exists on AWS 1:23:09