
Apptronik hired five senior leaders in April to take its Apollo humanoid to market, and none of them has physical AI in the title. Foxconn, which builds electronics, has a head of Physical AI. Before you choose a VP of Physical AI, decide which of those two jobs you're filling: building physical AI into a product, or putting robots to work inside your own operations.
Then hire someone who has already done that job. Both count as physical AI leadership, but the résumés they need have little in common.
Quick Read
- Before hiring, decide whether the VP will develop physical AI products or deploy robots within your own operations.
- While robots sit in a single-site pilot, an operations leader working with the vendor can usually carry the program.
- Robot makers often split the work, as Apptronik did with a VP of Software from Amazon and a services leader from Boston Dynamics.
- A robot program that kept running after the pilot is stronger evidence than demos or research credentials.
What Does a VP of Physical AI Own?
A VP of Physical AI owns the AI that lets machines act on their own in the real world, either inside a product the company sells or inside its own operations. At a robot maker, that may mean overseeing the models running on robots and the field data that improves them. At a manufacturer or logistics company, the responsibility shifts to getting robots into production and keeping them connected to existing operations.
The title is still settling. At SAP, Lukasz Ostrowski was head of Embodied AI and Robotics in January 2026. By November, SAP described him as Global Head of Physical AI and Robotics. People doing this work today often carry another title, such as VP of Robotics or Head of Autonomy, so look at the scope of each candidate's last job before the words on their business card.
The VP of Physical AI also sits close to two other jobs that get confused with it.
[Table 1]
When Is It Too Early to Hire a VP of Physical AI?
If your robots are in a pilot at one site and a vendor runs most of it, you probably don't need a VP yet. An operations leader with a strong vendor relationship can carry a single pilot.
The case for the hire gets stronger when a second site is coming, or when robot output starts showing up in commitments to customers. At that point someone has to own the program across sites, and the cost of a stalled rollout becomes visible in the numbers. Most manufacturers haven't reached that point yet.
In a Tata Consultancy Services survey of 300 manufacturing executives in North America and Europe, conducted in March and April 2026, only 9% had moved a pilot into operations, and 4% were running deployments across more than five facilities. Anupam Singhal, TCS's president of manufacturing, told Technology Magazine that people will keep providing “oversight and accountability” in these operations, especially where decisions affect worker safety.
Robot makers face a different version of the same question. Before the product reaches paying customers, the job is usually research and engineering, and a CTO or head of robotics often covers it. A VP focused on physical AI deployment makes more sense once robots are working at customer sites and someone has to keep them working.
How Are Companies Filling Physical AI Leadership Roles in 2026?
After closing a Series A that totaled more than $935 million, Apptronik announced the leaders who would take Apollo to market. Chirag Shah joined as VP of Software from Amazon, where he led multimodal AI work for Alexa+ and Kindle. Kevin Garell, who ran global services at Boston Dynamics, became SVP of Services and Support, and Daniel Chu, Waymo's former chief product officer, came in as CPO.
The useful part is how the hires divide the work. Apptronik brought in a software leader with consumer AI experience alongside an executive who had already run robot services at Boston Dynamics, keeping software development and field support with different people. I'd be surprised to find one candidate with Shah's background and Garell's at the same depth.
At Foxconn the problem looks different. Dr. Di Wang, its head of Physical AI, said in SAP's physical AI announcement that the company needs "physical systems that understand business context, not just motion commands." A manufacturer can buy capable robots. What it has to build itself is the link between those robots and its production plan. At Foxconn, that means robot inspections feeding straight into maintenance plans inside SAP's warehouse software.
Boston Dynamics shows both needs inside one group. It named Rohit Prasad CEO in October, bringing in Amazon's former AI chief to lead its physical AI push. Its parent, Hyundai Motor Group, plans to deploy 25,000 Atlas robots across Hyundai and Kia plants. Kia's CEO has said the robots should start work at Hyundai Motor Group Metaplant America in Georgia in 2028, with Kia's Georgia plant following in 2029. Prasad leads the robotics company, and Hyundai's manufacturing organization will still need clear accountability for how those robots perform on its production lines.
What Should You Look For in a VP of Physical AI?
You can start with proof that the candidate's work held up outside a lab. A demo shows talent, but it says little about whether a machine kept running through a night shift.
The evidence looks different for each mandate, so use the column that matches yours.
[Table 2]
A leader from consumer AI can succeed in a product role, as Apptronik is betting with Shah, provided someone beside them already knows field operations.
For operations roles, also ask about the people working beside the robots. That's where rollouts tend to slow down, and it has leadership needs of its own, covered in the five roles physical AI companies need in production.
How Can You Test a Candidate Before Making an Offer?
A visit to a working robot site can reveal more about a candidate than another round of interviews. At a pilot facility, or a customer site for robot makers, each finalist gets to see the machines in use and explain what they'd investigate first. Someone who has managed robots in production may notice operators repeatedly stepping in to correct a problem, or a machine slowing down the rest of the line. Their experience should come through in the questions they ask and the problems they recognize.
There's also the question of who can stop a deployment over safety concerns:
- Has the candidate ever delayed a launch because validation was incomplete?
- What happened when the plant wanted to proceed anyway?
Someone who has faced that situation should be able to explain how the decision was made and who was involved. The company needs to establish the same authority before the new VP takes over, particularly when robots will be working alongside employees.
Frequently Asked Questions About Hiring a VP of Physical AI
How Do I Choose a VP of Physical AI for My Business?
Start with the mandate. If the role builds physical AI into a product you sell, look for someone whose software runs on robots at customer sites today. If it runs robots inside your own plants or warehouses, look for someone who has taken a robot program from pilot into daily production.
Should a VP of Physical AI Come From Robotics or From AI?
Hire from whichever side your current team is weaker on. A company with deep hardware engineering can bring in AI leadership from software, while a company with good AI talent and little field experience needs someone who has kept robots running in production.
Who Should a VP of Physical AI Report To?
I'd tie the reporting line to the job. If the role builds physical AI into a product, it fits under the CTO or chief product officer. If it runs robots inside your operations, it belongs with the COO or the head of manufacturing, because that's where the results get measured.
Can an Internal Leader Become VP of Physical AI?
Yes, especially on the operations side. The manager who ran your robot pilot already knows your sites and the people running them. Before promoting them, check whether they handled the integration work themselves or whether the vendor did it for them.

