Hook
The financial disclosure forms landed with the quiet finality of a ledger entry. Emil Michael—the Pentagon's Chief Digital and AI Officer, the second person to ever hold that title—had been selling. First xAI stock in January, reportedly netting up to $24 million. Then, over the summer, a full exit from Perplexity. Two transactions. Two AI unicorns. One man whose day job is deciding how the U.S. Department of Defense adopts artificial intelligence.
The market read this as insider signal. I read it as a data problem.
Context
Let me establish the baseline facts before we interpret anything. Emil Michael is not a career bureaucrat. He came from Uber's executive suite, spent years in Silicon Valley's investment circles, and held equity positions in multiple AI companies before entering government service in April or May 2025. His portfolio included stakes in xAI—Elon Musk's foundation model lab—and Perplexity, the AI search engine that has been fighting a two-front war against Google's AI Overviews and OpenAI's ChatGPT Search.
The financial disclosures are public record. The transactions are not alleged; they are documented. What remains unclear is the why behind the timing, and that ambiguity is where the analytical value lives.
Core
Let me walk through the on-chain evidence—or in this case, the paper trail—with the same rigor I would apply to a DeFi protocol audit.
The xAI Exit: January 2025
The reported $24 million gain on xAI stock requires context. xAI's valuation trajectory in 2024-2025 was steep: from roughly $24 billion toward the hundreds of billions, driven by the Colossus compute cluster and Grok model iterations. If Michael acquired his stake in early rounds—say, 2023 or early 2024—the appreciation would be multiples, not percentages.
The January timing matters. Michael's formal entry into the CDAO role occurred around April-May 2025. A January sale sits in the window before assuming office, which is precisely when federal ethics rules pressure appointees to divest potentially conflicting assets. This looks like compliance, not conviction.
But here's the analytical wrinkle: if this were purely compliance-driven, why did the Perplexity sale happen in summer—after he was already in office?
The Perplexity Exit: Summer 2025
Perplexity's valuation trajectory in 2025 was equally dramatic: from roughly $3 billion toward $9 billion across multiple funding rounds. A summer exit means Michael sold into a rising market, capturing appreciation but also leaving future upside on the table.
Two competing narratives emerge:
- The Compliance Narrative: He was systematically cleaning his portfolio of conflict-prone assets, and the timing simply reflects when the ethics review process reached each position.
- The Signal Narrative: He had access to information about Perplexity's defense prospects—or lack thereof—and chose to exit before that information became public.
The data cannot distinguish between these. But the sequence is telling. If compliance were the sole driver, you would expect a single coordinated divestment event, not a staggered approach across months. The staggered pattern suggests either administrative lag or a more deliberate, asset-by-asset assessment.
The Competitive Read
From a competitive landscape perspective, the two exits tell different stories. xAI in early 2025 was riding a wave of compute advantage and Musk's deep government relationships through SpaceX. Perplexity, by contrast, was facing existential competitive pressure: Google's AI Overviews were eating search queries, and OpenAI's Search was targeting the same use case. The defense angle also diverges—xAI's infrastructure capabilities have obvious national security applications, while Perplexity's consumer search product has a less clear defense procurement path.
Michael's portfolio allocation—holding both—suggests he saw value in both the model layer and the application layer of AI. His exits suggest he saw different risk profiles.
Contrarian
Here is where I push back on the prevailing interpretation.
The market narrative treats these sales as insider pessimism. I would argue the opposite: the sales are the most bullish signal possible for the AI sector's institutional integration.
Think about it structurally. A senior Pentagon official held equity in two frontier AI companies. That alone confirms the defense-industrial complex is now formally intertwined with the AI startup ecosystem. The fact that he sold—whether for compliance or conviction—means the system is working as designed. Disclosure happened. Transactions were documented. The public can audit the paper trail.
The real risk is not that Michael sold. The real risk is that other officials hold AI equity and don't disclose it. This event normalizes scrutiny. It creates a template for what transparency looks like in the AI-military complex.
Correlation is not causation. The market assumes Michael's exit reflects company-specific knowledge. But the more parsimonious explanation is bureaucratic: he entered government, the ethics lawyers did their work, and assets were divested on a schedule determined by administrative review, not market timing.
Takeaway
The signal to track is not Michael's portfolio. It is the next 90 days.
Watch for three data points: whether the DoD Office of Inspector General or the Office of Government Ethics opens a formal review; whether Perplexity's next funding round closes at or above the $9 billion mark; and whether any other senior defense AI officials' disclosures reveal similar holdings.
If the OIG stays silent and Perplexity raises at a premium, the market will have its answer: this was compliance, not conviction. If the opposite occurs, the "insider signal" narrative gains empirical support.
Follow the disclosure forms, not the headlines. The data will tell you which story is real.