PHYSICAL AI & THE LEADERSHIP GAP THE COMPANIES WINNING IN PHYSICAL AI HAVE ONE THING IN COMMON: THEY'RE HIRING PEOPLE THEY'D NEVER HIRED BEFORE.

OUR EXPERTS Kara Ruskin Senior Client Partner Technology, Moses Zonana Senior Client Partner Technology, Alissa Moody Senior Client Partner Industrial Technology, Smriti Kangovi Managing Consultant Technology

Robotics. Industrial automation. AI/IoT. Industry 4.0. Different labels, different founding stories, different investor narratives but increasingly the same leadership challenge.

Leadership transitions across this sector are accelerating faster than the talent frameworks built to manage them. Boston Dynamics is searching for a new CEO after Robert Playter stepped down in February 2026 following thirty years at the company and six years as CEO. OpenAI's hardware leader also resigned in March 2026. These are businesses reaching an inflection point where the technology works, the market is real, and the leadership demands have fundamentally shifted.

The instinct most boards reach for in that moment is to hire the way technology companies have always hired: strong pedigree, recognizable companies, a pattern that feels safe. That is precisely where the problem starts. These are not software businesses that happen to have hardware. A deployment delay is not a sprint-planning issue, it is a supply chain, a certification, a customer contract. The cost of the wrong leadership decision is not a missed quarter; it is a safety incident, a recall, a regulatory unraveling.

Korn Ferry's latest research found that only 11% of talent leaders believe their executives are well-prepared to lead through the AI transition. In sectors where AI meets physical reality, that number should alarm every boardroom.

Over the past several weeks we pressure-tested that thesis against senior executives who have actually built, scaled, and commercialized in this space across large-scale industrial automation, precision manufacturing, and field robotics. What follows are nine convictions about this market and what they change about how leadership should be defined, sourced, and assessed. Physical AI is not a software problem with hardware attached. It is a contextualization problem, and contextualization is a leadership capability before it is a technical one.

Physical AI is real but unevenly distributed, and the market has not yet learned to tell the difference. Value is already being created today on factory floors, in supply chain operations, in warehouse automation, in product intelligence, and in life sciences manufacturing. Caterpillar has applied it across product lines and operations. Multiply Labs has used automation to turn cellular and gene-therapy manufacturing from an expensive, slow process into something materially faster and cheaper. That layer is deployed, and it is paying.

The humanoid layer is a different matter. The most candid assessment we heard put it at roughly 90% hype and 10% real value, with the remainder concentrated where the human form factor genuinely creates customer value. The ROI timeline we find most credible: humanoids reliably reducing defects on industrial lines within about three years, genuinely paying for themselves in performed functions after that, and moving meaningfully beyond industrial settings into retail and hospitality only in roughly a decade. The overall market still grows at a strong double-digit rate throughout. Both things are true at once, and boards need leaders who can hold them simultaneously.

The value has migrated from the model to the context around it. Model costs have collapsed and the model layer is commoditizing. Access to AI is no longer a differentiator. The scarce capability is contextualization: connecting AI to real equipment, real workflows, real-time data, simulation, digital twins, customer operations, and domain specific outcomes.

This has a hard operational consequence most companies underestimate. Years of investment in data lakes and traditional simulation are not sufficient. Extracting value on a factory floor requires real-time simulation and digital-twin capability wired directly into live operations. That is not an IT transformation. It is an operating-model transformation and it demands a different kind of executive to lead it. 

Commercialization, not invention, is now the binding constraint. The unsolved problem across this sector is profitability and production at scale. No company has yet nailed the balance between the physical product and the services layer around it. Remote contract manufacturing and hardware quality remain persistent sources of failure. Even the longest-standing, most recognized names are still working this out.

The discipline that matters is carrying something the full distance: idea to prototype to field testing to customer deployment to commercial service contract. That is a full-lifecycle muscle, and it is not one that most software leadership tracks ever build. The leadership challenge in this sector has shifted from invention to deployment and hiring frameworks have not caught up.

Simplification is a leadership capability and its absence is expensive. The single clearest example we heard: a major logistics operator initially scoped 27 distinct robotics applications for its fulfillment centers and wanted custom hardware for each. Pushed on the premise, that list collapsed to two core applications: pick-and-place, and towing items between bins and shelves. The correct architecture was standard hardware with the software and AI layer absorbing the variation.

The same logic applies to the hardware itself. If a task can be executed with six axes of freedom, building for twenty adds cost, reduces reliability, increases energy consumption, and frequently delivers nothing a customer will pay for. Our conviction: AI should reduce the need for custom hardware, not multiply it and a leader who instinctively customizes rather than standardizes will destroy margin at scale.

Safety, certification, and field reliability are strategic variables, not compliance line items. Safety remains one of the largest blockers to adoption, particularly for robots operating near people, even as sensor technology and software stacks improve. This has direct implications for how we assess leaders: product and growth fluency is not enough. Executives in this sector need genuine command of risk, certification pathways, field reliability, and customer trust.

It is also a competitive variable at the national level. US companies are more constrained by liability and risk considerations, while other markets have more latitude to experiment quickly. We expect that dynamic to shape where iteration speed accumulates versus where it does not.

Technical credibility cannot be substituted with general management strength. This is where we see boards make the most expensive mistake. The failure pattern is well documented in the sector: a company follows the traditional industrial playbook, prioritizes a strong general manager over a technically grounded leader, and absorbs real setbacks as a result.

Technical depth is genuinely hard to preserve as leaders move up day-to-day proximity to the technology erodes. But the leaders who hold it retain a specific, testable habit: they can sit with an engineer, listen, go away and study the problem, and return with real direction. That capability should be assessed directly. It rarely survives a conventional competency framework.

The complete profile is a unicorn, so the answer is adjacency mapping, not a harder search. The profile everyone is looking for combines AI, robotics, automation, end-market knowledge, business-model design, and operational execution. That person is close to nonexistent in the volumes this sector now requires. Continuing to search the same pool is not a strategy.

---

---

 

Our position is that the talent answer will increasingly come from outside robotics. For product leadership built on simplification, modularity, user experience, and scalable product systems, the relevant analogues are companies like Lego, Apple, and IKEA. For recurring revenue and services built on an installed base, GE Aviation and similar models are the reference point. For hands-on automation exposure, CNC machining, tooling, and large-scale manufacturing environments consistently produce people with meaningful automation fluency who have never carried a robotics title.

There is also a specific role we believe is systematically undervalued: the translator who sits between the customer and engineering, converting customer needs into engineering priorities, managing technical development, and surfacing commercial opportunities along the way. In field-deployed businesses, this is frequently the most important profile in the company, and it almost never appears on a board's slate.

The sector is recycling too much of the same talent. The same executives are circulating between the same companies, and the pattern is reinforcing rather than solving the capability gap. We see a real opportunity for leaders who are not traditional robotics executives, but who already understand scale, simplification, recurring revenue, productization, customer deployment, and operational discipline and who can learn the domain quickly. Domain knowledge is acquirable. Operating instinct at scale is much less so.

Talent supply in this sector is a structural and geopolitical variable, not a recruiting detail. The US continues to face an engineering talent shortage, addressable primarily through workforce development and continued inbound talent from abroad. Other markets, most notably China and India, are producing more engineers than they can currently absorb, creating a fundamentally different supply-demand dynamic. We expect data and iteration speed to accumulate where experimentation is least constrained, and compute and chip advantage to remain where it currently sits. Boards building multi-year capability plans should treat this as a planning input, not background noise.

WHAT THIS CHANGES Taken together, these convictions point to a different definition of the role than most Physical AI briefs currently contain. We now push clients to assess for:

• Contextualization judgment the ability to connect AI capability to live equipment, workflows, and customer economics, rather than fluency with the model layer itself.
• Full-lifecycle commercialization demonstrated movement of a product from prototype through field testing to a paying commercial contract, not just launch experience.
• A simplification instinct evidence of collapsing complexity and standardization rather than proliferating custom solutions.
• Retained technical credibility tested directly, not inferred from an engineering degree or an early-career title.
• Safety, certification, and field-reliability command treated as core commercial judgment, not delegated risk management.
• Boundary-crossing history: candidates who have held hardware and software realities simultaneously rather than sequentially.

And it changes where we look. Sourcing strategies anchored exclusively on robotics and AI companies are structurally incapable of filling these roles at the volume the market now needs. Adjacent industries with comparable operating complexity consumer product systems, installed-base services, precision manufacturing are where a meaningful share of the answer sits.

Supply-chain and reshoring leadership is emerging as a related but distinct capability boards should assess directly: demonstrated experience localizing or diversifying component sourcing and manufacturing footprint under real time and cost pressure, not just overseeing an existing domestic supply chain.

WHAT WE ARE WATCHING Market structure in humanoids is consolidating around a small group with real manufacturing capability and ecosystem depth at companies like Boston Dynamics, Tesla, Figure AI, Agility Robotics, and Apptronik. Agility Robotics is among those closest to a genuinely defensible humanoid application, backed by real commercial deployments rather than demos alone.

Outside humanoids, the established incumbents FANUC, ABB, KUKA, and Universal Robots remain highly durable. FANUC in particular remains deeply entrenched on installed base. Notably, even these entrenched leaders are choosing to partner at the AI layer rather than build it themselves. At NVIDIA's GTC 2026 event, all four confirmed they are integrating NVIDIA Omniverse and Isaac simulation frameworks into their virtual commissioning and digital-twin offerings. That is itself a signal for how boards should think about buy-versus build leadership decisions.

On demand, we believe deglobalization is becoming a primary automation driver in its own right. As companies build production capacity closer to home or within specific regions, advanced automation becomes the only viable path to cost competitiveness at smaller regional scale. Reshoring, data centers, aerospace and defense, space, life sciences, mining, and agriculture are all pulling in the same direction. One Americas automation business we spoke with has grown at roughly 22% year over year for twelve consecutive quarters. Automation is no longer only a technology trend, it is a strategic response to deglobalization, labor constraints, productivity pressure, and supply-chain resilience.

This is compounded by a shifting policy backdrop. Recent trade and national-security measures affecting cross-border robotics supply chains are accelerating conversations boards were already having about where components are sourced, where manufacturing sits, and how resilient that footprint is to disruption. Regardless of how individual policies evolve, the direction of travel is consistent: supply-chain sovereignty is moving from a procurement conversation to a board-level one.

That has a direct talent consequence. Leadership teams increasingly need executives who have actually led domestic reshoring, dual-sourcing, or supply-chain localization efforts, not as a compliance exercise, but as a core commercial capability. This is a profile that, until recently, sat outside the standard robotics or Physical AI leadership brief, and one we believe boards should begin assessing for explicitly.

THE CONVERSATION BOARDS SHOULD BE HAVING The uncomfortable implication of all of this is that boards and investors may be using the wrong scorecards entirely over-indexing on software pedigree and recognizable technology brands at precisely the moment those signals predict the least. The companies getting this right are not searching for who they have hired before. They are building clarity on who they have never hired, and why they need them now. That is the hardest conversation in the room. And it is also the only one worth having. What's the one leadership quality your current hiring process isn't actually assessing for?

#PhysicalAI #ExecutiveSearch #LeadershipHiring #Robotics #AITransformation #IndustrialAutomation #FutureOfLeadership #TalentStrategy