Jeff Bezos’ “physical AI” lab, Project Prometheus, is said to be nearing a $10 billion funding round at a roughly $38 billion valuation, according to a report by the Financial Times. For ERP users, the significance goes beyond venture capital: it points to a future in which AI is used not just to generate content or insights, but to improve how operations actually run across manufacturing, logistics, warehousing, packaging, and other ERP-intensive environments.

Project Prometheus is co-led by Jeff Bezos and Vik Bajaj, a former Google X scientist, with Bezos taking an operational co-CEO role, his first since stepping down from Amazon. The company, which launched in November 2025 with $6.2 billion in initial funding, is focused on so-called “physical AI,” a category aimed at building models that understand physics, interact with real environments, and support work across engineering, manufacturing, robotics, aerospace, and other operational domains.

In other words, this is not primarily about generating text, images, or code on demand through large language models (LLMs); it is about creating AI that can perceive, decide, and help execute in settings where machines, materials, assets, and workflows move through time and space.

Over the past two years, most enterprise AI conversations have focused on copilots, assistants, and productivity gains for knowledge workers. Physical AI points in a different direction: into factories, warehouses, supply chains, maintenance operations, and packaging environments where ERP systems already play a central role in planning, tracking, and control.

How Physical AI Connects ERP to Operations

For enterprise technology leaders, that is the important shift. Generative AI changed how users search, summarize, and create content, but physical AI aims to change how companies run plants, warehouses, maintenance operations, packaging lines, and supply networks.

That makes Project Prometheus relevant far beyond the AI model market, because its success would depend on whether it can convert industrial data and operational context into repeatable business outcomes.

Also, physical AI does not replace ERP; it adds a more responsive layer between enterprise plans and what is actually happening in operations.

In practice, that means AI could help manufacturers adjust production based on machine conditions, improve maintenance timing, and detect quality issues earlier.

In warehouses and logistics environments, it could support faster picking, smarter replenishment, and better response to disruption.

In packaging and line operations, it could help equipment deal with variability that traditional automation often struggles to handle.

The early impact is likely to be strongest in industries where ERP is already deeply embedded in day-to-day execution. Manufacturing, logistics, distribution, consumer goods, and asset-intensive sectors all depend on structured planning but operate in conditions that change constantly. That makes them the most natural fit for AI designed to connect planning with action.

From Productivity AI to Operational AI

For ERP users, the investment in Project Prometheus is another sign that AI is moving beyond office productivity and into the operational core of the business. The first enterprise AI wave centered on copilots, search, and content generation. The next may be judged by whether it can improve throughput, reduce downtime, strengthen quality, and make supply chains more resilient.

That shift has implications for ERP systems themselves. As physical AI matures, ERP platforms will need to connect more effectively with plant systems, warehouse technologies, robotics platforms, and industrial data streams. The question for enterprise software vendors and customers alike is whether ERP can remain the trusted process backbone while newer AI-driven systems handle more of the real-time execution layer.

It also raises a broader issue for organizations: data readiness. Physical AI depends less on public data and more on operational data collected from assets, workflows, sensors, and machines. For ERP-heavy businesses, that puts a premium on connecting transactional records with live operational signals in a usable and governed way.

In that sense, the Prometheus funding story is not just about one AI lab. It is a signal that investors see the next big enterprise opportunity in connecting digital systems to physical operations. For companies running complex ERP estates, that makes this less a technology curiosity than a roadmap issue.