Keep pulling the thread on Satya Nadella.
Microsoft's valuation has grown by a factor of 10 since Satya Nadella became CEO in 2014.
Satya Nadella's strategy of following developers and startups was a key driver behind Microsoft's acquisition of GitHub.
Microsoft is actively developing AI-generated, personalized user interfaces.
Satya Nadella envisions a future of knowledge work involving specialized, IDE-like applications for professions such as an "accountant IDE" or a "lawyer IDE" to manage interactions with AI agents.
Microsoft is experimenting with a "mission control" concept in GitHub Copilot where developers can deploy autonomous agents on different branches, with the future IDE centered on triaging the resulting pull requests.
Unlike the dot-com bubble's speculative build-out, Microsoft's current AI infrastructure build-out is supply-constrained and struggling to keep up with existing demand.
Microsoft does not have a GPU utilization problem, stating that every GPU it can deploy is sold out and in use.
The primary bottleneck in Microsoft's AI infrastructure expansion is the availability of "powered up shells" for data centers, which is constrained by long lead times for land, power, and building permits.
Satya Nadella predicts that the future of a corporation will be defined by its ownership of a proprietary foundation model that captures the company's unique tacit knowledge.
Stripe is training a proprietary "payments foundation model" using transaction data from its network to improve capabilities like fraud detection.
Microsoft has a project called "NL Web" aimed at creating a natural language interface for every merchant's product catalog, allowing AI agents to perform deep, reasoning-based searches.
Microsoft's strategy for products like GitHub Copilot is to use an ensemble of models and an intelligent router that selects the most appropriate model for a given task based on its complexity.