The venture capital perspective on hardware has completely flipped. Previously seen as a difficult and capital-intensive area to be avoided, hardware is now considered one of the few remaining defensible moats. Physical products like WHOOP and Eight Sleep create customer lock-in that software alone cannot.

Andromeda Surgical accelerates its development by using a commodity robotic arm from KUKA. This strategy allows them to focus resources on their core IP—the AI and software autonomy layer—rather than sinking capital and time into custom hardware development.

Instead of building its final passenger jet, Boom first developed a smaller, sub-scale prototype to prove its Mach 2.2 technology. This startup-like, sequential approach proves the core concept at a much lower cost, making the capital-intensive project more manageable and fundable.

Contrary to the belief that hardware is inherently capital-intensive, Monumental's founder argues their biggest expense is salaries for high-quality talent, much like a software startup. The cost of the robots is manageable and their payback time is good, challenging typical VC perceptions of the business model.

Dinoosh says Clone bet that frontier models and robot data would improve without it building the brain. He argues model half-life is six to twelve months, so the company focused on defensible muscle hardware and plans to post-train open checkpoints with its own robot data later.

The venture capital industry has reversed its historical aversion to hardware. In an AI-driven market where software moats are shrinking, the difficulty and capital intensity of building physical products like robots are now seen as a source of strong, long-term defensibility.

Investing in deeply technical hardware, like Cerebras's wafer-scale chip, requires a degree of ignorance about the true difficulty. Unlike software, where a core technical insight gets you 80% of the way, in hardware it's only 2%. The rest is a brutal, multi-year battle against physics and complex supply chains that experts might avoid entirely.

The VC landscape has dramatically shifted, with Physical AI funding reaching $45 billion in H1 2026, up from under $5 billion in 2023. This counters the long-held VC belief that hardware startups are too risky, margin-thin, and susceptible to copying by competitors.

Dinoosh says Clone’s design can be produced at a tenth of the cost of motor-actuated humanoids because it uses a different supply chain. Lon notes the gears those rivals need are expensive and sourced from the same bottlenecked suppliers.