Over the past few years, there has been a fundamental shift in conversations with global CEOs and CXOs. One question keeps surfacing: How do we scale intelligence, rather than just scaling output? We have seen AI evolve rapidly from simple Q&A assistants to predictive tools to generative AI and more. Now, we are entering the next phase: AI as embodied intelligence. What once sat at the periphery of business conversations has now become the center of the enterprise strategy.
Cognitive machines are not science fiction anymore; they are entering real enterprise environments like Google DeepMind's Gemini Robotics pilots and Foxconn's humanoid testing for Apple manufacturing. For instance, UBTech Robotics has deployed dozens of Walker S1 humanoid robots in Zeekr’s EV manufacturing plant, where they collaborate to lift materials and handle factory logistics alongside human workers.
Aside from industrial applications, humanoids are also entering personalized, individuals spaces. At the Mobile World Congress 2026, Honor unveiled its first humanoid robot, which will focus on three core scenarios: shopping assistance, workplace inspections, and especially supportive companionship. X Technologies' NEO is marketed as the first "consumer-ready humanoid robot designed to transform life at home". Each NEO can have conversations, navigate around the home to provide help where needed, and with continued use, learn and develop new skills.
And as they do, leaders must start thinking beyond automation toward true collaboration between humans and machines. This is where humanoids come into play. What if your workforce included collaborators who learn from every interaction, anticipate challenges, and improve continuously, just like your top engineers? While CEOs have engineered digital transformations, they are now building the next era of enterprise, powered by cognitive machines.
The Evolution from Automation to Cognition
The early generation of industrial robots followed a script. Since they were designed for precision and repetition, they followed a set of instructions in a controlled environment. However, in the face of variability, they weren’t able to perform tasks with the required levels of speed and efficiency. When conditions changed, they required human intervention or further programming. For example, Gartner forecasts that fewer than 20 organizations are expected to bring humanoid robots to full production in manufacturing and supply chain settings by 2028, highlighting challenges that persist for real-world deployments.
Advances in large-scale neural networks, multimodal AI, edge computing, and sensor technologies are enabling a new class of machines capable of operating in complex, real-world environments. These cognitive systems can perceive their surroundings, interpret context, learn from data, reason, and adapt their actions in real time, even in unfamiliar settings. More than simply executing predefined instructions, they can respond dynamically to variability. In manufacturing environments, this capability is already being tested at scale. For example, BMW deployed Figure AI’s humanoid robot at its Spartanburg factory, where it handled over 90,000 components and supported the production of more than 30,000 BMW X3 vehicles during a 10-month pilot. BMW Group Plant Leipzig is running a pilot where AEON, the humanoid robot, is being introduced into production in Germany for the first time, AEON will support people, and not replace them, by taking on repetitive tasks, delivering materials to the line, and navigating around obstacles.
These advances are beneficial for enterprise leaders, because cognitive machines:
- Can operate in unstructured environments
- Can collaborate safely with humans
- Reduce reprogramming costs
- Improve with data
- Scale across use cases
- Can become an active participant and partner with existing talent to deliver faster and with greater efficiency
Engineering the Learning Brain
Engineering cognitive machines demands embodied intelligence designed for real-world complexity. A humanoid, at its core, is a physical system built to operate in environments originally structured for human interaction. To function effectively in these environments, machines must interpret multiple sensory signals simultaneously: vision, language, motion, and spatial cues. Multimodal AI enables this contextual reasoning beyond single-channel data. For example, Apple’s Ferret model shows how multimodal AI can combine vision and language to better understand context in real-world environments.
In the case of humanoids, learning is powered by continuous machine learning loops: ingesting environmental data, processing it through neural models, and refining performance through reinforcement signals. While initial training requires human guidance, the performance of these autonomous systems increasingly improves through self-optimization. Learning and evolving within a controlled environment are foundational. The true measure of cognitive systems is determined by their ability to adapt to the unpredictable complexity businesses face every day.
Mastering Adaptation in Dynamic Environments
When we speak of adapting to chaos, we are referring to the enterprise reality defined by variability, ambiguity, and constant change. Supply chain shifts, demand fluctuations, operational conditions evolving without notice, and more. Cognitive humanoids are designed to respond in real time. They detect early signs of disruption and quickly assess the best course of action by evaluating multiple 'what-if' scenarios, adjusting in real-time to keep operations moving. For example, Amazon has begun testing Agility Robotics’ Digit humanoid robots in its fulfillment centers to automate repetitive warehouse tasks such as moving containers between logistics stations.
This capability is powered by hybrid AI architectures, where structured, routine rules handle daily operations (supervised training), and anomaly-detection systems identify unexpected patterns (unsupervised training). Such adaptability is especially critical in high-stakes industries such as healthcare and financial markets. In healthcare, intelligent systems can dynamically reprioritize workflows during sudden surges in patient demand. In financial operations, they can detect unusual market movements and adjust execution strategies in real-time to manage risk. Now Cloud AI platforms like Google Cloud are increasingly used for real-time anomaly detection in financial data.
As leaders, the more relevant question is not whether disruption will occur but how resilient and shockproof our operations are when it does. Cognitive machines strengthen this resilience by adapting in real time. They function not merely as tools but as active collaborators within the enterprise.
The Power of Human-Machine Collaboration
We can utilize the full potential of humanoids when we use them as copilots. While humanoids become executors, performing tasks tirelessly, humans remain the creative minds behind them as dreamers and innovators. IDC forecasts that by 2026, around 40% of job roles in Global 2000 organizations will involve direct collaboration with AI agents. Meanwhile, ecosystems such as Microsoft Copilot, Meta’s open AI frameworks, and Apple’s on-device intelligence signal a shift toward deeply embedded human-machine interaction. For example, Mercedes‑Benz is piloting humanoid robots from Apptronik in its manufacturing facilities, where the Apollo robots assist with repetitive tasks such as moving components to the production line and performing quality checks alongside human workers. In March 2026, Amazon acquired Rivr, a Swiss robotics company developing machines for “doorstep delivery.”” This technology when working alongside delivery associates can improve safety outcomes and the overall customer experience, particularly in the last steps of the delivery process.
Humanoids are emerging as productivity multipliers in the enterprise system. Enabled by natural language commands (such as “Pass the tool”) and emotional AI that senses stress, the collaboration between humans and humanoids can feel seamless. With humanoids, you can free your brilliant minds for creativity and bigger-picture strategy, while they handle routine and high-frequency tasks with consistency. For true collaboration between humans and AI, trust is essential. That is where governance comes into play.
Ethical Engineering and Governance Imperatives
As AI becomes deeply embedded across our operations at scale, governance must scale with equal intensity to address challenges like data bias, cyber risks, and workforce uncertainty. Emerging regulatory frameworks such as the EU AI Act and evolving global standards make clear that responsible deployment is a foundational requirement. Safeguards will help ensure reliability and trust:
Audit logging: Log every AI decision and allowing human override
- Privacy-enhancing techniques: Enable systems to learn locally without sharing sensitive data
- Fairness evaluations: Conduct regular bias checks across diverse user groups
Once these safeguards are in place, the full potential of cognitive machines can be realized responsibly. In the era of cognitive machines, trust and governance will define your competitive moat. With that foundation in place, the next step is to define the road ahead.
The Roadmap Ahead
As AI becomes more widely adopted across industries, the journey ahead can be broken into three practical stages.
Short-term: Scale and affordability.
Faster deployment is possible when more affordable humanoids, powered by open-source AI, make it cost-effective to introduce them into factories, warehouses, and operations. These systems will increasingly rely on hybrid edge cloud architectures, where real-time inference occurs at the edge while training, monitoring, and orchestration remain centralized in the cloud.
Mid-term: Smarter coordination.
Greater efficiency and teaming can occur when multiple humanoids work together in real-time, automatically dividing tasks among themselves, while humans supervise to ensure precision, safety, and control. Ecosystem convergence and interoperability will become critical, as scalable value depends on seamless integration across data, platforms, and workflows.
Long-term: Human-machine partnership at the core.
Humans and machines learn from each other, improving decisions, execution, and overall business direction in real-time, resulting in joint learning and improvement. Over time, human-machine partnerships will become core to enterprise design. Industry forecasts suggest the humanoid robot market could grow significantly in the coming decade to over $11 billion by 2030, as AI and robotics technologies mature. Leaders across industries are already making significant investments to build this future. Those who move early will shape the next decade of business.
The Time to Lead
The era of cognitive machines and humanoids is already here, redefining enterprise. To enable successful deployment, leadership conversations must adapt and change. Industry research, including insights from McKinsey & Company, suggests that enterprises embedding AI deeply into operations are positioned to outperform peers on growth and efficiency. The future belongs to enterprises that move beyond adoption and systematically engineer these cognitive systems into core business.
Sources:
Reference 1 // Reference 2 // Reference 3 // Reference 4 // Reference 5
