Is it possible that a full-size pickup is to be a “seasonal” item, changing it like a wardrobe? Anduril cofounder Palmer Luckey said at an Andreessen Horowitz founders event that AI-driven production will drive the cost of producing physical objects as close to the floor as possible, and the sticker shock applied to cars and houses. “I really do believe that in our lifetimes you’ll be able to go buy something that’s like a Ford F-150 for $1,000,” Luckey said. “The cost of extracting and transforming it will go to near zero, and we’re going to compete the margins way down. It’s just not that crazy.”
Luckey applied the concept to end-of-life economics, where he talked of vehicles being cheap enough, as well as recyclable enough, to be treated as short-run assets. At the time of the season, he said, “I bet you’ll be able to recycle [a car] with 90% efficiency at the end of the season,” he said, imagining consumers choosing a “summer car” and swapping it out as casually as a new appliance.
The assertion falls in a sector where cost has taken a reverse direction in terms of affordability. However, the closer conflict is not one of AI optimism and consumer pricing; but that which software is able to squeeze and that which is physically needed by factories. “components are not expensive,” Luckey insisted. “It’s the transformation and the regulation that have made it really expensive.” It is that focus on change that is shifting the fastest today in the manufacturing playbook, with the cost center increasingly becoming the step-by-step-join-inspect-rework-logistical-changeover sequence, as opposed to the bill of materials.
Internet car manufacturers are already assaulting that cost of transformation with radical parts consolidation. A good example is of high-pressure aluminum die-casting of very large structural components, a change that eliminates assemblies of stamped components and welds in favor of a few large castings. The equipment itself represents the industrial volume of the bet: the presses produce 6,000 to 9,000 tonnes of locking force, which is programmed to inject large shots of aluminum in milliseconds with tight dimensional specifications. In ideal implementation, the number of parts will reduce the number of joints, the number of workstations and the amount of material that will need to be handled and the amount of time that will be used to connect and verify the same structure in dozens or hundreds of operations.
But gigacasting also demonstrates the difficulty of achieving “near zero.” The castings systems need furnaces, vacuum systems, temperature control, trimming, and intensive inspection infrastructure; the production line is more automated and more data-hungry simultaneously. There is also the workforce problem: reduced operators rejoining steps in manuals, increased technicians operating networks, robotics, process controls, and quality systems which must remain stable throughout long production cycles. Even under a situation where the factory is able, uptime and repeatability is the larger limitation since a mistake in a immense structural casting can bundle cost rapidly.
This is where AI in manufacturing ceases to be a buzzword and “transforms” into a number of extremely specific levers. Plant automation practitioners refer to AI-enabled machines as a means to transform thousands of process variables into decisions with the least intervention, predictive maintenance and real-time monitoring can be used to reduce unexpected downtime. The greatest savings are usually served by the avoidance of waste, scrap, and line stoppage-costs, which creep on without a murmur within the transformation itself that Luckey is talking about.
Digital twins extend the same rationale to a greater level. An analytics-led digital twin of a car, part, or assembly line enables an engineering team to model virtually, and inject factory and field information into design and manufacturing environments. In the case of EV programs, that feedback loop can be used to bring structural design, thermal behavior and manufacturing constraints into agreement prior to physical locking of decisions with physical tooling.
The amount of the $1,000 pickup is provocative, yet the fact is less dramatic and already in process: factories are restructuring around fewer steps, more automation, stricter inspection, and richer data. When hardware becomes much more affordable it will not be due to the fact that the materials became gratis; it will be that the industry has learnt how to waste less time and less touchpoints to transform materials into something that ships.