What Does Zebra’s Acquisition of Fetch Robotics Amount to?

Zebra Technologies, a global logistics technology developer with a market value of US$25 billion, has acquired the final 95% of Fetch Robotics after initially investing in 5% of the company in 2019. The final valuation for Fetch Robotics was US$305 million, close to thirty times the current yearly revenue of Fetch, which is around US$10 million. This valuation is largely in line with previous acquisitions of AMR vendors. The comparative case is Teradyne’s acquisition of Mobile Industrial Robots (MiR) in 2018 for US$272 million. MIR recently posted US$42 million in revenue for 2020, assisted by the expansive ecosystem of distributors and system integrators utilized by Teradyne’s Industrial Automation division.

While MIR and Fetch Robotics were started only one year apart, the revenue gap is striking. Importantly, Fetch can utilize the resources of Zebra Technologies, which dominates the bar-code scanning market. While MIR is focused primarily on production logistics as a critical market and emphasizes small fleet sizes per deployment, Fetch has centered around warehouses, distribution, and fulfillment centers. Fetch also has a much more comprehensive sensor modality and much more powerful computing architecture. Besides this, they are one of the few autonomous mobile robot (AMR) companies to be cloud-dependent. This approach, built on always being online, brings the benefits of improved computing capacity but is yet to gain trust among many traditional customers who are skeptical about sending data beyond the premises. The cloud-centric approach of Fetch is also enabling Robotics-as-a-Service, where customers can pay based on usage.

Fetch and MiR differ in many ways, but they both follow a similar philosophy in that their navigation is built on dynamic rerouting. While many automated guided vehicles (AGVs) focus on teach and repeat route creation, Fetch’s systems can use their simultaneous localization and mapping (SLAM) algorithms to find the best reroute depending on obstacles and traffic. As far as Fetch and MiR is concerned, this emphasis on ‘dynamic mapping’ is critical to the concept of an AMR. They have emphasized this in being critical to the definition of what an AMR is. They have been influential in setting the terms and regulations for modern AMR safety, as emphasized by the recently released ANSI/RIA R15.08.

Many other vendors, such as Balyo, AGILOX, and Seegrid, take a different definition of autonomy. They argue that autonomy is based on lack of external infrastructure, or that a robot can move without needing markers, magnets, or external sensors to perceive and move about. So, from this perspective, even if a robot is running on a virtualized linear route, it remains autonomous and an AMR. These systems, generally with much higher payloads, run on teach-and-repeat setups. Their autonomous robots can be controlled to change routes by a fleet management system, but they cannot dynamically reroute.

For years, the key to progress for AMRs has been more dynamic movement and flexibility. Based on this trend, companies like Fetch, MIR, and Otto Motors are pursuing the right course. They will increasingly seek to improve their perception capabilities. This will include the ability to segment the environment between humans, vehicles, and so forth (panoptic segmentation), and gather ever more data for cloud-based operations and analytics.

But the AMR industry will not necessarily develop in a linear fashion. Companies like Seegrid, while interested in improving the flexibility of their systems, are confident that their main customer base, namely large manufacturers, prefer technical redundancy over flexibility. There may be something to this argument based on Amazon’s rollout of new AGVs and AMRs. Some commentators pointed out that Amazon’s systems are not cutting edge as they run on pre-determined routes and do not utilize SLAM. But the same could have been said for Kiva Systems back in 2012. After deploying 350,000 old-school QR-code enabled AGVs, Amazon wants to scale up quickly. Dynamic robots may maximize autonomy and flexibility, but the more prominent players seem arguably more interested in deploying large-scale solutions with a higher degree of predictability and technical redundancy. If the competition is between autonomy maximalists that to create the most responsive robot and those looking to see robots as a form of extended infrastructure, the latter is currently taking the spoils.

This being said, Fetch’s pedigree and eventual acquisition is a further validation of highly autonomous mobile robots and portends well for the coming decade. Taking an expansive definition of AMRs to include any system that does not require physical infrastructure, ABI Research estimates that the material handling space will grow from 15,000 shipments in 2020 to 170,000 shipments in 2025.

It is also a great boon for Zebra Technologies, which, like Teradyne, is acquiring new technologies to provide new revenue streams to complement their dominance in mature technology fields. Ultimately, it continues the story of consolidation as individual companies are drawn into wider industrial collectives. Zebra can learn a lot from Teradyne’s success, but they can also benefit from their unique capabilities from Fetch. While Teradyne has an expansive fleet, it is highly fragmented with very small numbers of robots sold for each installation. In order to get real efficiencies from robots, it is important to scale. Fetch has already built relationships with the likes of DHL, and going forward, Zebra would do well to focus on a select number of large installations where maintenance costs and service lifecycle management challenges can be controlled. To further ameliorate the challenges of scalability, they could work with third parties like Ricoh (a previous partner of Fetch) to rollout their solution comprehensively on a global basis.