Zach Weinberg on why pharma tariffs miss the point and why AI won't cure disease anytime soon

Summary

Zach Weinberg, who works in early-stage biotech and drug discovery, makes two arguments that cut against the current Washington consensus on pharmaceuticals and AI: pharma tariffs are targeting the wrong part of the value chain, and AI is nowhere near the cure-for-disease moment that frontier lab fundraising pitches imply.

Pharma tariffs miss the point

The Trump administration's signal that pharmaceuticals are next for tariffs gets the strategic logic backwards, Weinberg argues. By the time a drug reaches manufacturing, the hard work is done. Discovery and development — the years of research, failed compounds, and capital risk that run into hundreds of millions of dollars per program — are where America's competitive advantage actually sits. Drug manufacturing, by contrast, is largely a quality-control process that doesn't require large workforces and is cheaper to run offshore. Tariffing finished-goods manufacturing raises costs for American patients without protecting anything strategically valuable. The part worth protecting, Weinberg says, is drug discovery and drug development, and tariffs on the last mile do nothing for that.

Why government funding in biology is non-negotiable

The reason private capital can't replace NIH funding isn't ideology — it's patent law. You cannot patent a biological discovery. Understanding how a cell-signaling pathway works, or what is actually driving Alzheimer's progression, produces no IP and no revenue. Without a business model, there is no private investor. Pharma and biotech companies can patent the drug they design to intervene in a known biological process, which is why venture and pharma capital flows into drug development but not into the upstream basic science that makes development possible in the first place.

Alzheimer's is his sharpest example. Researchers can observe that certain proteins accumulate in the brains of Alzheimer's patients, but whether those proteins are causing the disease or are a byproduct of something else entirely remains unknown. Without that biological understanding, there is no rational target for a drug program. Weinberg's view is that without government or philanthropic funding for that open-ended lab work, it simply won't happen — and the drug pipeline for diseases like Alzheimer's will eventually run dry.

His case for growing the NIH budget is indirect but pointed: a wealthier America funds more science. The current NIH budget is roughly $30 billion. If the US economy were ten times larger, that number could be $300 billion or more, which Weinberg frames as one of the most concrete reasons to care about economic growth beyond its immediate effects on wages and prices.

AI in biotech: useful, not transformative yet

Weinberg's website doesn't mention AI, a deliberate choice that reflects his view of where the technology actually sits. His summary is blunt: a useful tool that may improve the efficiency of a few steps in the process, but not a panacea. The reason is structural. Making a drug is roughly a 40-step process, and AI can help with some of those steps — but the binding constraint remains biology, not compute. If you don't know what is causing a disease, no model can tell you what to target. AlphaFold solving protein folding was a genuine scientific advance, but biotech markets barely moved because investors understood that knowing a protein's structure doesn't tell you what it does in disease or how to drug it safely. The hype around LLMs curing disease, Weinberg implies, tends to surface most visibly when frontier AI companies need to raise capital — and the biotech market's muted response is the more honest signal.