Can you shed some light on why they would fire 4k American workers while concurrently aplenty for 2.5k H1B visas? Just seems like they are destroying American workers to bring over foreigners. Maybe there is more to it but it needs to be explored.
The model-cost convergence is the quiet story. Everyone argues about benchmark crowns, but the real enterprise unlock is when capable tool-using models get cheap enough to be used messily, repeatedly, and inside boring workflows. That is where software spend starts looking less like seats and more like compute-backed labor.
The convergence of these three trends is the most definitive evidence we’ve seen yet of the Zero-Marginal-Cost Trap hitting the enterprise layer, Chamath.
When Grok 4.5, GPT-5.6, and Muse Spark 1.1 drop input token costs down to $1 per million, "agentic intelligence" ceases to be a proprietary corporate moat—it becomes a subsidized commodity utility.
This explains the structural margin pressure hitting Microsoft. When an enterprise optimizes a legacy business model purely around mass, frictionless distribution (like Game Pass or standard enterprise seats) while the underlying cost of compute is collapsing to zero, margins inevitably bleed out. Starbucks building its own AI to replace legacy enterprise apps is just the first wave: sophisticated clients are realizing they don't need to rent an intermediary's wrapper when the raw commodity is this cheap.
The boardroom lesson is stark: if your enterprise value is built entirely on automated execution, you are actively automating your margins out of existence. True corporate alpha must be anchored in the absolute scarcity of sovereign, human-verified relationship loops (Rh) and ring-fenced proprietary truth.
This Wednesday in The Unclonable Boardroom, I’m releasing the mathematical foundations of the Trust Capital Audit Index (TCAI) to give boards a concrete tool to measure this exact exposure.
Given your new role on the Fed task force on Productivity and Jobs, I’m curious: how are you and the task force thinking about measuring true economic productivity when traditional volume-based output (Y) becomes hyper-inflated by zero-cost synthetic data? Are we tracking a shift from a volume economy to a sovereignty economy?
Can you shed some light on why they would fire 4k American workers while concurrently aplenty for 2.5k H1B visas? Just seems like they are destroying American workers to bring over foreigners. Maybe there is more to it but it needs to be explored.
The model-cost convergence is the quiet story. Everyone argues about benchmark crowns, but the real enterprise unlock is when capable tool-using models get cheap enough to be used messily, repeatedly, and inside boring workflows. That is where software spend starts looking less like seats and more like compute-backed labor.
The convergence of these three trends is the most definitive evidence we’ve seen yet of the Zero-Marginal-Cost Trap hitting the enterprise layer, Chamath.
When Grok 4.5, GPT-5.6, and Muse Spark 1.1 drop input token costs down to $1 per million, "agentic intelligence" ceases to be a proprietary corporate moat—it becomes a subsidized commodity utility.
This explains the structural margin pressure hitting Microsoft. When an enterprise optimizes a legacy business model purely around mass, frictionless distribution (like Game Pass or standard enterprise seats) while the underlying cost of compute is collapsing to zero, margins inevitably bleed out. Starbucks building its own AI to replace legacy enterprise apps is just the first wave: sophisticated clients are realizing they don't need to rent an intermediary's wrapper when the raw commodity is this cheap.
The boardroom lesson is stark: if your enterprise value is built entirely on automated execution, you are actively automating your margins out of existence. True corporate alpha must be anchored in the absolute scarcity of sovereign, human-verified relationship loops (Rh) and ring-fenced proprietary truth.
This Wednesday in The Unclonable Boardroom, I’m releasing the mathematical foundations of the Trust Capital Audit Index (TCAI) to give boards a concrete tool to measure this exact exposure.
Given your new role on the Fed task force on Productivity and Jobs, I’m curious: how are you and the task force thinking about measuring true economic productivity when traditional volume-based output (Y) becomes hyper-inflated by zero-cost synthetic data? Are we tracking a shift from a volume economy to a sovereignty economy?
Microsoft's stock price might be taking a beating right now, but its 39% net profit margin is phenomenal compared to the 10%-15% industry average.