Former DoD Advisor on How Silicon Valley is Rewiring the U.S. Military

The conversation opens with a framing device the advisor returns to throughout: dictatorships are simultaneously enormously strong and enormously weak. They control the apparatus of the state, but their fundamental illegitimacy creates deep internal distrust that cascades through their institutions, including their militaries. This framing is applied to both Iran and China.

On Iran and the Strait of Hormuz, the advisor argues that 'winning' is politically defined. Degrading Iranian military capability and reopening the Strait would constitute a strategic win, but the challenge is the hybridized ideology of 'Red Shiaism' — a fusion of Marxism and martyrdom theology articulated by Shariati — which allows adversaries to interpret their own destruction as victory. He draws parallels to Japan's kamikaze culture in World War II, arguing that fanaticism requires an extreme quantum of force to overcome, and even then, regime change requires an alternative power structure for people to defect to, which Iran currently lacks due to IRGC control of half the economy.

The advisor discusses the competing military doctrines of Maxwell Taylor (flexible, proportional response) and Eisenhower (massive retaliation/nuclear deterrence), concluding that both are necessary. A purely maximalist nuclear deterrence posture would leave no gradations of response, while flexible power allows enforcement of smaller infringements — Tomahawk missiles, HIMARS, and similar systems being examples.

On China, the advisor expresses confidence that China will eventually fall, comparing it to the Soviet Union, which appeared monolithic until its sudden collapse. He notes Xi Jinping's fundamental illegitimacy creates paranoia — the Chinese equivalent of the Chairman of the Joint Chiefs disappeared despite being a childhood friend of Xi's. He argues the U.S. has rebuilt significant clandestine capabilities to exploit this internal distrust, and that every member of the Chinese standing committee has relatives in the U.S., making America the 'bugout point' even for China's elite.

On Taiwan, the advisor suggests a military invasion is less likely than a political accommodation, particularly if the KMT wins elections, and that China may achieve its goals without firing a bullet. However, he warns that a military takeover of Taiwan would likely not stop there, given historical grievances with Japan and others.

On the Ukraine conflict as a laboratory for modern warfare, the advisor highlights how commercial viability of drones transformed them from niche tools into central instruments of war. He notes the extraordinary iteration speed of Ukrainian drones — roughly 50 generations in three years — enabled by commercial supply chains and garage-level development. This underscores his broader argument that the commercial world now directly drives military capability.

On procurement and defense industrial base reform, the advisor is critical of one-year Congressional appropriations, continuing resolutions that prohibit new program starts, and the absence of multi-year contracting authority for ordnance and critical systems. He advocates for repointing a portion of FFRDC budgets toward commercial industry, creating multi-year authorities across the stack, and allocating a higher percentage of the defense budget to higher-risk, higher-reward innovation — analogizing the needed investment to the Manhattan Project's scale relative to GDP.

On the co-option of American universities by Hamas and the Muslim Brotherhood, the advisor references FBI transcripts from a Philadelphia hotel room post-Oslo Accords, where Hamas leaders allegedly laid out a 20-year plan to shift Western opinion through academic institutions — a plan he argues succeeded.

Finally, on AI, the advisor expresses concern about adversarial poisoning of AI training data through fake academic papers and fabricated medical conditions, warning that if language models are incorporated into military decision-making loops and their foundational data is corrupted, the models could become 'towers of babel' — dangerously wrong in ways that are hard to detect.