The story of a Formula One engineer trading the high‑octane world of Red Bull Racing for the gritty reality of factory floors captures a broader shift in engineering talent. After experiencing the pinnacle of aerodynamic performance, Bercan Kilic sensed a desire to apply his problem‑solving skills to challenges that affect everyday life—namely, how goods are produced at scale. His decision reflects a growing trend where specialists from elite motorsport, aerospace, or high‑frequency trading are redirecting their expertise toward solving systemic inefficiencies in manufacturing and logistics. The $55 million seed round secured by his Munich‑based startup, microagi, not only marks the largest early‑stage financing ever raised by a German company but also signals that deep‑tech investors are increasingly willing to back ambitious visions that bridge cutting‑edge AI with physical automation. This infusion of capital arrives at a moment when the robotics sector is poised for a step‑change, driven by converging forces such as aging workforces, geopolitical pressure to reshore supply chains, and the declining cost of sensors and compute.
Microagi’s core proposition is deceptively simple yet technologically nuanced: rather than manufacturing robots or inventing foundational AI models from scratch, the company focuses on the data‑centric problem of teaching existing robotic systems to perform specific tasks in real‑world settings. By equipping human workers with cameras and sensor‑laden gloves, microagi captures fine‑grained demonstrations of activities ranging from assembly line operations to food processing. These recordings are then fed into partner‑provided robotics models, which are fine‑tuned to replicate the demonstrated behaviors within a client’s factory environment. This approach sidesteps the enormous upfront capital and lengthy development cycles associated with building proprietary hardware, allowing manufacturers to adopt automation more rapidly. Crucially, it also creates a feedback loop where each deployment generates additional data that further refines the models, creating a virtuous cycle of improvement that is difficult for competitors to replicate without a similarly extensive data collection network.
The timing of microagi’s funding round aligns with a palpable surge of investor interest in what is often termed “physical AI” or “embodied intelligence.” After years of excitement surrounding large language models and generative AI, capital is now flowing toward ventures that translate algorithmic advances into tangible actions in the physical world. Several macro‑economic factors are amplifying this shift. Persistent labor shortages in Europe and North America—exacerbated by demographic trends—are pushing manufacturers to seek alternatives to human labor for repetitive or physically demanding tasks. Simultaneously, rising geopolitical tensions and the fragility exposed by recent global supply‑chain disruptions have intensified the incentive to reshore production, a move that is only economically viable if automation can keep labor costs competitive. Finally, the steady decline in the price of high‑performance computing, advanced sensors, and lightweight actuators has lowered the barrier to entry for deploying sophisticated robots in settings that were previously considered too costly or complex.
Quantitative evidence underscores the urgency of adopting automation at scale. According to the International Federation of Robotics, China accounted for 54 % of all factory robots installed worldwide in 2024, deploying a staggering 295,000 units compared with just 34,200 in the United States. This disparity highlights not only China’s aggressive push toward manufacturing automation but also the risk that Western economies could fall further behind in productivity and cost competitiveness if they do not accelerate their own adoption curves. For European manufacturers, the stakes are particularly high: the region’s median age has risen to nearly 45 years, and projections indicate the workforce could shrink by almost 19 million people by 2050. In such a context, automation is not merely an option for improving margins; it is becoming a prerequisite for maintaining industrial viability. Microagi’s focus on enabling factories to train robots quickly and affordably directly addresses this structural challenge, offering a pathway for European firms to close the automation gap without requiring massive upfront investments in proprietary robotics platforms.
One of the most profound bottlenecks hindering the broader deployment of capable robots is the scarcity of high‑quality, task‑specific training data. While large language models have been trained on corpora containing hundreds of billions of text tokens harvested from the internet, no comparable repository exists for robotics. UC Berkeley roboticist Ken Goldberg famously described this mismatch as a “100,000‑year data gap,” emphasizing that the amount of demonstrable human experience available to train robots is minuscule relative to what fuels today’s AI chatbots. Microagi’s consumer‑facing arm, shift, attempts to bridge this divide by paying individuals across 15 countries to record themselves performing everyday physical chores—such as dishwashing, floor mopping, or laundry folding—in exchange for services like free apartment cleanings or private chef sessions. The resulting dataset, which now comprises contributions from more than 20,000 participants, provides a rich source of varied human motion patterns that can be used to pre‑train or fine‑tune robotics models, thereby accelerating the learning curve for machines tasked with similar operations in industrial contexts.
From a business‑model perspective, microagi operates at the intersection of data services and AI enablement. On the B2B side, it partners with manufacturers who seek to deploy robots for specific, repeatable tasks; microagi captures the necessary demonstrations, adapts third‑party robotics foundations to the client’s environment, and supports the rollout. On the consumer side, shift functions as a data‑collection engine that monetizes the gathered footage by licensing it to AI labs and robotics firms developing foundational “brains” for machines. This dual‑track strategy creates multiple revenue streams while ensuring a steady inflow of diverse, real‑world motion data that keeps the company’s adaptation pipeline robust. Notably, microagi refrains from developing its own foundational models, instead relying on specialized partners such as Physical Intelligence, Skild AI, and Generalist AI. By focusing on the adaptation layer—where customer‑specific nuances are encoded—the startup can remain agile, avoid the massive compute costs associated with training giant models from scratch, and still deliver tailored solutions that meet the exacting demands of factory floors.
Assessing the technological readiness of today’s robotics ecosystem invites a useful analogy to the evolution of language models. Just as the GPT‑2 release demonstrated that scaling data and compute could yield predictable performance gains, the robotics community is currently searching for the equivalent inflection point where increased data, better algorithms, and cheaper hardware translate into reliable, general‑purpose machine competence. Kilic characterizes the present state as akin to a “GPT‑2 moment” for robotics: promising signs exist, but the technology is not yet consistently useful across broad applications. He anticipates that the field is approaching a “GPT‑3.5 moment,” wherein the combination of sufficient task‑specific data, refined model architectures, and mature deployment tooling will enable robots to handle a wide variety of factory jobs with minimal bespoke engineering. Achieving this threshold would unlock the potential for widespread adoption, turning robots from niche, high‑maintenance assets into plug‑and‑play tools comparable to today’s industrial PLCs or CNC machines.
The founding team’s background provides a compelling illustration of how cross‑disciplinary expertise can accelerate innovation in hard tech. Alongside Kilic’s Formula One aerodynamics pedigree, the cofounders include a former Mercedes F1 engineer, a researcher from the Alan Turing Institute, an RWTH Aachen specialist in mechanical systems, and a serial entrepreneur with a track record of bringing deep‑tech ideas to market. This blend of high‑performance engineering rigor, theoretical AI insight, practical mechanical design, and commercial acumen equips microagi to tackle both the technical nuances of robot adaptation and the go‑to‑market challenges inherent in selling to conservative manufacturing clients. Their collective experience in environments where milliseconds and millimeters matter—such as Formula One—translates into a meticulous approach to sensor calibration, latency reduction, and reliability testing, all of which are critical when robots operate alongside human workers in shared workspaces.
The $55 million seed round, led by Hummingbird with participation from Northzone, LocalGlobe, Village Global, and redalpine, represents more than a financial milestone; it is a validation signal from the venture capital community that Europe can produce deep‑tech champions capable of competing on a global stage. Hummingbird’s track record—including early stakes in AI‑focused ventures like Lovable, Kraken, and Etched—suggests a pattern of backing founders who combine technical depth with audacious scaling ambitions. The size of the round also reflects the capital intensity required to sustain the compute‑heavy workflow of fine‑tuning robotics models at scale, as well as the need to incentivize and manage a geographically dispersed data‑collection network through shift. For European policymakers eager to foster homegrown champions in strategic sectors like automation and AI, microagi’s fundraising success offers a case study in how ambitious vision, paired with demonstrable traction, can attract significant late‑stage‑style capital even at the seed phase.
Concerns about job displacement frequently accompany discussions of workplace automation, yet microagi’s leadership frames the issue through a lens of labor supplementation rather than substitution. Kilic argues that in many Western factories, the impending retirement of seasoned workers is not being matched by an inflow of new talent, creating a structural shortage that threatens output capacity. In this view, robots serve as a means to preserve and even augment productive capacity by taking over tasks that are difficult to staff, thereby allowing human employees to focus on higher‑value activities such as process oversight, quality control, or creative problem‑solving. Empirical evidence from industries that have already embraced collaborative robots supports this narrative: firms often report improved employee satisfaction and reduced injury rates when robots handle ergonomically taxing or dangerous operations. Nonetheless, the transition will require proactive reskilling initiatives and thoughtful change‑management strategies to ensure that the workforce benefits from, rather than merely endures, the shift toward greater automation.
Looking ahead, microagi has articulated an ambitious roadmap that hinges on three primary uses of its fresh capital: scaling the compute infrastructure needed to train ever‑larger robotics models, expanding the shift data‑collection footprint to capture an even broader array of human motions, and deepening its presence in the United States to tap into one of the world’s largest manufacturing markets. The company’s cofounders have set a daring internal benchmark: within five years, they aim to have microagi’s technology running on tens of millions of robots globally, with a failure condition defined as deploying fewer than 20‑30 million units. Achieving such scale would necessitate not only technical excellence but also mastery of regulatory compliance, partnership ecosystems, and after‑sales support across diverse industrial verticals. If realized, this deployment footprint could meaningfully shift global productivity curves, particularly in regions grappling with demographic headwinds, and potentially redefine the competitive balance between established manufacturing hubs and emerging economies.
For stakeholders navigating this rapidly evolving landscape, several actionable insights emerge. Investors should scrutinize startups that prioritize data moats over hardware differentiation, as the ability to gather and leverage task‑specific human demonstrations may prove more defensible than investing in yet another robot manipulator. Manufacturers seeking to embark on automation journeys would be well advised to begin with well‑defined, repeatable use cases, capture high‑quality demonstration data internally or via partners like shift, and adopt a phased rollout that measures key performance indicators such as cycle time, error rate, and worker satisfaction before scaling. Policymakers, meanwhile, ought to consider incentivizing the creation of open, annotated datasets for robotics—similar to public image corpora like ImageNet—to lower the entry barrier for smaller firms and accelerate collective progress. Ultimately, the convergence of AI, robotics, and demographic trends presents a once‑in‑a‑decade opportunity to rebuild industrial systems that are more resilient, efficient, and inclusive; the winners will be those who treat data as the foundational raw material and human expertise as the essential guide for teaching machines to work.