Scarcity-First Engineering
Engineering isn't the perfect build — it's overwhelming the problem for a fixed budget
- Difficulty
- Moderate
- Time to result
- ~months to results
- Steps
- 5
- Confidence
- 70%
Lonsdale reframes engineering as a scarcity problem, not a capability problem. Anyone can build a bridge that never falls if given a billion dollars; the point of engineering is to build it for fifty million and have it never fall just as much. Applied to hardware and defense, this means questioning exquisite specs: a vehicle could carry all thirty-seven requested capabilities, or you could build a thousand of them with ten capabilities for the same cost and simply swarm the adversary. Because you always operate under a fixed resource budget, the winning move is often overwhelming mass of cheap-enough units rather than a handful of gold-plated platforms. The corollary is that advanced manufacturing becomes strategically critical, since the side that can produce cheap mass fastest wins the exchange.
Origin
Extracted from Lonsdale's defense-building work at 8VC and companies like Saronic and Overland AI, where he contrasts traditional prime-contractor specs with attritable-mass designs, and ties the argument back to why advanced manufacturing capacity matters.
Core principles
- 01Anyone can build the perfect thing with unlimited money; engineering is doing it cheaply
- 02Scarcity is always the real constraint, so design to a resource budget
- 03Mass of cheap-enough units can beat a few exquisite ones
- 04Do not gold-plate specs no one actually needs
- 05Overwhelming quantity is itself a capability
How to run it
- 1
Fix the budget first
State the total resources you are willing to spend before enumerating desired capabilities.
Pro tip Budget is the design constraint, not an afterthought once specs are set.
- 2
Separate needs from padding
Distinguish the handful of genuinely required capabilities from spec inflation nobody actually needs.
Watch out Long requirement documents are often written to justify one expensive solution.
- 3
Run the mass-versus-exquisite comparison
At equal cost, compare one fully-loaded unit against many simpler units and estimate which wins the real engagement.
Pro tip A thousand ten-capability vehicles can overwhelm one thirty-seven-capability vehicle.
- 4
Design to overwhelm
Choose the configuration that swamps the problem within budget rather than the one with the best spec sheet.
Watch out Superior specs lose if the other side fields far more units.
- 5
Build the manufacturing base
Invest in advanced manufacturing so producing cheap mass is actually achievable at scale and speed.
Pro tip Whoever manufactures cheap mass fastest wins the attrition exchange.
In the wild
Lonsdale describes Saronic building hundreds of smaller autonomous Navy vessels, and Overland AI aiming to build thousands of autonomous ground vehicles. Rather than meeting a traditional spec of thirty-seven capabilities and heavy armor, the design question becomes whether the same budget buys a thousand ten-capability units that can swarm and overwhelm the adversary, accepting that many will be lost.
→ Attritable mass that overwhelms costlier, more capable single platforms.
Common mistakes
Gold-plating the spec sheet
Loading every capability onto one unit maximizes the brochure but produces something too expensive and too few to win.
Ignoring the manufacturing constraint
A cheap-mass design is useless if you cannot actually manufacture the volume fast enough to field it.
Is it for you?
Best for
Hardware and defense builders deciding between a few premium units and many attritable ones.
Not ideal for
Domains where a single failure is catastrophic and quantity cannot substitute for reliability.
From the transcript
“The point of engineering is not to build a billion dollars. It's how do you build it for 50 million dollars, never fall down just…”
“You could give it all 37 of those capabilities, or you could build a thousand of them for the same cost with like 10 capabilities.”
From the episode
How To Win The War Of The Future - Joe Lonsdale - #934
Joe Lonsdale