CRISPR was announced in 2012. Emmanuelle Charpentier and Jennifer Doudna described a molecular scissors system that could edit any gene in any organism. The headlines said we would cure dozens of diseases within years. Fourteen years later, the first CRISPR-based therapy for chronic Hepatitis B is just entering serious clinical development. That gap is not a failure. It is the system working exactly as designed, and understanding why matters more than complaining about it.
Drug development runs 20 years end-to-end. That number comes from René Kuijten at EQT Life Sciences, and it is not a pessimistic estimate. It is an engineering reality. You discover a target, you validate it in cells, you validate it in animals, you run Phase I safety trials, Phase II efficacy trials, Phase III large-scale trials, then regulatory review. Each stage exists because the previous one cannot tell you what the next one will. Biology does not compress on demand.
The Part Nobody Funds
Here is where the system actually breaks down. The 5-to-7-year middle phase, after basic discovery and before late-stage trials, is where most promising science dies. Not because the science is wrong. Because it is too de-risked for venture capital and too applied for academic grants. Kuijten calls this the phase where value creation is greatest and the work is most complex. It is also chronically underfunded. Dementia research sits here right now, mirroring where oncology was 20 years ago: real mechanistic understanding, limited therapeutic options, persistent funding gaps in the translation layer.
Metformin is the counterargument people reach for. Isolated from Galega officinalis in the early 20th century, introduced in the UK in the late 1950s, it still anchors type 2 diabetes treatment today. The argument goes: old drugs work fine, so why chase novelty? That is a fair point about the value of incremental science. But metformin took 6 decades to reach standard of care, and we only understand its full mechanism now. That timeline is not a model to celebrate. It is a baseline to beat.
CRISPR's journey illustrates the Gartner hype cycle with painful precision. Peak of inflated expectations in 2012 to 2015. Investment pullbacks when delivery to living cells proved brutal and off-target DNA cuts created safety concerns. Then the slow, unglamorous climb: better delivery vectors, improved base editors, refined guide RNA design. Thousands of researchers across hundreds of labs, most of them unnamed in any press release, grinding through the painstaking work. That is how this science moves.
Where the Speedup Is Real
The Gladstone Institute's Geneformer model is doing something genuinely different. It predicted novel heart disease regulators that decades of traditional research had missed, compressing target identification from years to hours. That is not hype. That is a measurable reduction in one specific phase of the pipeline. Christina Theodoris's team is not claiming to cure heart disease. They are claiming to find better targets faster, which means clinical trials with higher success probability. That is the right framing.
AI cannot compress Phase III trials. It cannot replace the 10,000-patient study that tells you whether a drug kills people at scale. Biological validation has irreducible time costs, and anyone selling you a 5-year drug development timeline is selling you something else entirely.
The ask is specific: sustained NIH funding for mid-stage translation research, patient capital from biotech investors willing to hold through the 5-to-7-year valley, and AI deployment targeted at target identification and trial design, not marketed as a shortcut past the hard biology. The 20-year pipeline shrinks when we stop defunding its middle and start engineering its front end. Not before.
