Claims

OpenAI's strategy is not just about building chips or foundries, but about viewing AI infrastructure holistically as a critical commodity that includes energy, data centers, chip design, and networks—and requires solving the entire ecosystem together rather than optimizing individual pieces.

Training and inference compute costs have followed a multiplicative growth pattern: GPT-3 cost $100 million with 175 billion parameters, while GPT-4 cost $400 million (4x) but with nearly 10x the parameters, suggesting costs continue to grow with each subsequent model generation.

This is probably the best time to start a company since the internet, at least, and maybe in the history of technology, because AI will enable increasingly remarkable capabilities each year and the greatest companies and most impactful products are created at technological inflection points like this.

Shoshana Zuboff's 'Surveillance Capitalism' framework, released in 2018-2019, was terrific at explaining the forces driving changes in social media that others had only partially described.

Microsoft and other major tech companies have initially disappointed with their AI implementations, with co-pilot being a limited initial offering, but now the agentic vision of AI is showing real value in terms of employee augmentation and customer intimacy.

Salesforce's model is to move from the hype cycle to the value cycle by delivering concrete productivity gains and demonstrating clear ROI; this approach of showing actual results differentiates them from companies still in the hype phase.

Super-scaler companies (large data center operators like AWS, Google Cloud, Alibaba) competing for dominance are investing hundreds of billions in data centers, driving down commodity prices for compute, which benefits software companies like Salesforce that can run applications across multiple providers at very low cost.

Robot workers deploying in restaurants represent a concrete example of digital labor taking hold in real-world commercial applications.

Microsoft disappointed many companies because while it started with co-pilot, it did not deliver the value it promised initially, but now real value is emerging in agentic applications.

Salesforce's large customers are actively deploying agentic AI applications, demonstrating that value is moving beyond hype into implementation phase.

Designing an advanced chip requires paying for software tools (hundreds of thousands or low millions of dollars) rather than billions, creating much lower barriers to entry in chip design relative to manufacturing.

Wolfram Alpha functions as an Oracle for blockchain smart contracts, providing real-world data (current weather, stock prices, etc.) that smart contracts need to execute conditions, and this connection is how Wolfram Research became interested in blockchain technology and led to collaboration with projects like Cardano

Microsoft's approach to platform control was relatively benign compared to Apple's: Bill Gates said the OS should make less money than apps, whereas Apple charges 30% and kills major apps that don't play ball.

There is significant pent-up developer energy that has been suppressed by closed platforms over the past decade, and the right blockchain application could release this energy in a wave of innovation similar to early Twitter's developer excitement.

Stack Overflow's traffic is down 40% after its API was shut down, likely because large language models trained on Stack Overflow's open-source contributions, devaluing both the platform and the programmers who contributed there—this may be a canary in the coal mine for all content-dependent AI.

Open source software broke the network effect monopoly that concentrated all software power into the OS vendor, enabling efficient business creation that would have been impossible under OS-vendor monopoly conditions.

No major apps will build for Vision Pro because developers have been 'so screwed over' by Apple's 30% take rate and control that they avoid app store dependence.