The gap between AI's operational capacity and its actual deployment value is widening faster than the gap between Grok and GPT's voting preferences. Grok's AI overwhelmingly recommends Republican candidates while OpenAI's GPT favors Democrats across more than 90 percent of nearly 500 election races, but this political divergence matters less than what it signals: models trained on different data and incentives produce different outputs, and nobody is surprised anymore. Meanwhile, the infrastructure for running AI has become almost trivially cheap and accessible. Microsoft open-sourced Agensh, a multi-agent harness scaling to 1,024 agents without central orchestration. Anthropic released sandbox-runtime as open-source OS-level sandboxing for agents. Cognition open-sourced the Devin Vulnerability Fixes plugin to patch PHP and Node.js deployments. Block built a voice model that runs locally on a MacBook M4 Max. Free coding agents proliferate through OpenCode Zen, Gemini's free quota, and others. The constraint is no longer building or running models. It is knowing what to build them for and whether users will actually pay for the result. Pete Steinberger spent ten million dollars on AI tokens in a single month at OpenAI while building OpenClaw, an AI digital assistant. The PHP Meltdown showed that AI tools can uncover years-old vulnerabilities and accelerate attacks using them. TechCrunch's running list of failed AI projects now includes Apple's repeatedly delayed Siri AI and OpenAI's messy super app launch. The industry is littered with agents designed for text messages, families, travel, and work, none of which have proven they justify their token costs or the attention they demand. Employers still post job listings demanding "AI experience" without clarity on what that means. The real story is not whether AI works. It is whether anyone has figured out how to make money from it beyond selling compute to people like Steinberger who are still burning capital at venture-scale burn rates while looking for product-market fit.
Sloane Duvall