Top 3 SPAC Targets – AI Robotics

Sometimes a sector has been hot on the private investment side for some time but it takes a high-profile SPAC deal getting a real pop on announcement to launch a trend.

That pop definitively came with Churchill Capital XI’s (NASDAQ:CCXI) $3.4 billion deal with Agility Robotics in June. The announced tie-up with the humanoid robot maker spiked Churchill XI’s shares above $19 and it still trades at about $14.65 a month later.

That reaction is perhaps surprising considering that robotics was a dicey sector for SPACs in the past cycle. Out of five business combinations completed with robotics targets in 2021-2022, there was one clear winner – SVF 3’s deal with warehouse automation firm Symbotic (NASDAQ:SYM), which trades consistently above $40 and hit a high of $87.88 in the past year.

The other four, however, include one target that agreed to be bought out for $1.40 per share two years after closing its SPAC deal the other three have either gone bankrupt or now trade at the equivalent of less than $1 after accounting for share splits. As a result, in spite of Symbotic’s success, robotics de-SPACs have had a median share price return of just $0.89.

At the time of their deals, these three lagging de-SPACs were each developing different types of human-piloted robotics – Sarcos Robotics (NASDAQ:PDYN) was building exoskeletons to increase worker strength, Nautilus (NASDAQ:KITT) was building submersibles to do underwater maintenance and Vicarious Surgical created robotic arms that could allow doctors to do surgical procedures remotely.

The problem with these approaches is that they would necessarily encounter some adoption resistance and could only scale with as many individual workers chose to use them as tools. AI changes all of that.

Now that powerful AI systems are available to guide robotics platforms themselves, the latest generation of robots promise to actually replace some workers entirely rather than add one more expensive tool to their belt.

Bright Machines

Perhaps no company is more emblematic of AI’s impact on this space than Bright Machines.

The San Francisco-based company has designed its robotics platforms not only to be powered by AI, but to have the specialized use case of building AI data centers themselves. It estimates its systems can lower overall data center buildout costs by -30%, with a +40% faster time to revenue and +15% greater initial reliability for the site.

In fact, Bright Machines can deploy its production lines directly on the site of new data centers and actively assemble and install the server racks itself. This notion of robots building robots would of course not be a popular business model for Will Smith’s character in “iRobot” (2004). But, it places Bright Machines even deeper into the story of being among the picks and shovels plays that are meeting the booming demand for AI datacenter growth.

It is perhaps for this reason that Bright Machines has become the second-most funded private robotics startup to date, according to New Market Pitch. It has raised $385 million, with its latest round being a $126 million Series C.

That round closed in June 2024, putting Bright Machines on a path to likely needing more capital soon. Considering how many recent SPAC deals have involved backing a data center firm about to launch a major infrastructure buildout with all of the execution risk that entails, how about one that will be building them autonomously and holding the IP on the robots building them to boot?

Path Robotics

Path Robotics foresees a somewhat less specific deployment, but nonetheless one that will be in growing demand.

Its Obsidian physical AI platform uses both assembly line configurations and mobile robotic dog-like platforms specialized in welding tasks. Through their own training these machines have welded about 10 million inches of contact points and now get it right on the first pass 97% of the time.

Because this learning is housed in the Obsidian AI brain itself, Path can continue to adapt to different needs as different physical robotics platforms come available. The need for that work is also expected to increase fast in the coming years.

Path estimates the average trained welder in the US is only actively doing torch work for up to 12% of their work day, while a robot would have much higher uptime. Furthermore, the average age of a US welder is 55 and thus a large portion of this workforce will be retiring in the coming years.

As it works to get more of its robots deployed, Path is offering clients arrangements that force them to spend $0 on capex for the systems, presumably in exchange for subscription or licensing fees. That could be a small price to pay, especially if the Path robots hit their targets of 4x the average human welder’s productivity and -30% the overall cost.

Covariant

While Path measures its AI brain’s existing work in the millions of inches, Covariant’s internal systems have been trained on billions of physical tasks.

At the moment, its robots have been tasked with warehouse product picking work, which may be low-hanging fruit compared to with work what Path and Bright Machines have put on their robots’ shoulders. But, Covariant’s systems beat out their human counterparts by even larger margins.

The accuracy of human warehouse pickers varies based on the level of technology they are using, with pickers using paper-based organizational systems hitting an average accuracy range of 90% to 96%, according to engineering group MTLI. That can rise to as high as 99.7% with better technology on hand, but Covariant’s robots clocked in at 99.96% accuracy with each arm moving 500 items per hour in a deployment with logistics firm Capacity.

That difference of about 0.26% is massively important for a company moving millions of products per year, especially when each mistake could lead to a product return or even worse outcomes in pharmaceutical applications.

Covariant’s last capital raise was a $75 million Series C extension that closed in April 2023 to help specifically meet growing customer orders for its systems and it brought its total private funding to $222 million.

At the time of that Series C, the company had about 300 robots at work in 15 countries and was coming off of a year of 6x growth. It has likely only become more thirsty for capital since.