Hadrian CEO Chris Power argues US contract manufacturing has no existing capacity, so companies must build about 50 percent more than future sales capacity before any contract arrives.
Power says every CapEx bet stays company-killing until roughly $50bn in scale, and puts Hadrian's company-death risk still near 80 percent.
Power says spending about 30 percent more on flexibility beats ordering equipment that later proves wrong, because CapEx and long engineering projects are one-way doors.
Hadrian builds factory capacity before contracts, betting on flexibility
Hadrian is betting that American manufacturing can only scale if capacity exists before contracts do. Standardized, reusable stations let most capital move between programs, which turns a one-way bet into a hedged one. The risk stays high, so the approach depends on paying extra for flexibility and on financing built for heavy capital.
Takeaways
Build capacity before any contract arrives. US contract manufacturing has no existing capacity, so a company must build roughly 50 percent more than its future sales capacity upfront. Power compares this to AWS, which Amazon ran internally before selling it to outside customers such as Stripe.
Modular stations make capital flexible and hedgeable. Standardized GPU-style production stations let about 80 percent of CapEx move between programs. On a hypothetical $300m casting facility with a two-year build, Power says paying 20 percent more for full flexibility beats betting on one demand forecast.
Capex bets stay company-killing until huge scale. CapEx is binary for large companies, with about $1bn deployed before contracts, and every bet stays company-killing until roughly $50bn in scale. Power still puts Hadrian's death risk near 80 percent, falling only when growth slows to about 20 percent a year.
Treat capex as a one-way door and overbuild. Power spends about 30 percent more to bake in hedges rather than order equipment that proves wrong. He pays $25m for foundations instead of $15m to save four months of layout simulation, and overinvests in electrical drops.
Plan factories for failure, not perfect yield. Power rejects assuming a 90 percent yield and builds slack capacity, like a freeway off-ramp. When about 40 machines from one vendor began to rust, he pressed its president for a four-week fix and the vendor flew in 30 engineers.
Heavy capex needs prepayment or long-duration credit. Only customer prepayment, as NASA gave early SpaceX, or long-duration infrastructure credit makes CapEx-heavy businesses work. Hadrian was underwater on financing costs until its late Series B, and Power says SaaS accounting at firms like Databricks differs.
Growth demands fast hiring and small teams. Hadrian has about 725 employees and plans about 600 more, while its pipeline grows by $2bn every quarter. Engineering leaders run programs of 30 to 40 people, and Power keeps only about two cultural values.
Power's leadership rules
Screen candidates hard. Power screened about 20 a day in his first twelve weeks.
A CEO of a complex company must be second-best at every function.
Tap a name to go there · tap a line to see why they are linked
On this episode 2
Show 1
Write Ulysses contracts while clear-headed, because sleep deprivation and hype degrade judgement.
Maximize product velocity rather than fixing release dates, as Ramp does.