A bet on taking AI beyond the screen

Jeff Bezos has placed one of the largest bets in the artificial intelligence race outside the chatbot arena. Prometheus, the “physical AI” startup he co-founded with Vik Bajaj, raised $12 billion in a round that values the company at roughly $41 billion. The size of the round is enormous, but the deeper editorial point is that capital is moving toward AI systems designed to affect engineering, manufacturing, chips, medicines, and real-world industrial processes.

According to TechCrunch, Prometheus wants to build an “artificial general engineer”: software capable of automating parts of the design and manufacturing process for complex physical systems, from aircraft engines to pharmaceutical compounds. GeekWire, summarizing Bezos and Bajaj’s public interview with CNBC, reported that the company has about 150 employees across San Francisco, London, and Zurich, and that it has not yet announced widely available products or a firm launch date.

What makes Prometheus different

Prometheus does not appear to be positioning itself as a simple productivity tool for engineers. Its ambition is to turn processes that today require years of simulation, prototypes, testing, multidisciplinary teams, and manufacturing coordination into end-to-end AI problems. Bajaj used the example of an aircraft engine, a system that can take a decade or more to design and produce. If that cycle can be meaningfully shortened, the impact would extend beyond technology into industrial costs, defense, energy, transportation, and health care.

The funding round also shows who wants to be close to that possible shift. TechCrunch reported participation from Bezos, JPMorgan Chase, Goldman Sachs, and BlackRock, among others. GeekWire also mentioned DST Global and Arch Venture Partners, citing Axios. The signal is clear: major investors are not treating physical AI as a lab curiosity, but as a potential infrastructure layer comparable to cloud computing, semiconductors, or industrial automation.

The uncomfortable part: automating engineering is not free

The promise comes with major questions. Automating text, code, or images is already difficult; automating physical engineering involves materials, safety, regulation, validation, failures, and legal responsibility. A model proposing a molecule, turbine, chip, or medical device cannot be wrong with the same tolerance as a chatbot drafting an email. In the physical world, mistakes cost money, delays, and in some sectors, lives.

There is also a labor question. Bezos has argued that AI-driven productivity could raise living standards and create a kind of “labor scarcity,” where demand for people exceeds supply. But that view coexists with another possibility: that intermediate technical tasks, documentation, preliminary design, repetitive simulations, iterative testing, and manufacturing data analysis could be automated or centralized inside global platforms. The outcome will depend on how these tools are adopted: as copilots that expand human teams or as systems designed to replace entire functions.

Why this matters for Puerto Rico

For Puerto Rico, this story does not imply an immediate impact or confirmed Prometheus operations on the island. But it does touch a strategic point. Puerto Rico’s 2025 manufacturing profile from the Department of Economic Development and Commerce says manufacturing represented 44.2% of Puerto Rico’s GDP in fiscal year 2024. The agency’s 2025 pharmaceutical profile says the pharmaceutical industry contributes about 30% of GDP and that pharmaceutical exports reached $48.328 billion, equal to 74% of the island’s total exports.

That means the physical AI conversation is not abstract for Puerto Rico. The island has real exposure to advanced manufacturing, pharmaceuticals, medical devices, regulatory compliance, and operations where design, validation, quality, and documentation are core parts of the business. If platforms like Prometheus succeed in accelerating the development of molecules, materials, equipment, or industrial processes, local operations could benefit from efficiency, but they could also face pressure to reorganize functions, reduce routine tasks, or depend more heavily on systems developed outside Puerto Rico.

The risk is not only job loss; it is also the loss of internal decision-making capacity. If design, validation, and optimization knowledge becomes locked inside AI platforms controlled by a small number of global companies, Puerto Rico could remain a user of imported technology instead of participating as an adapter, auditor, and creator of solutions for its own industrial ecosystem.

Prometheus still has to prove that its vision works beyond the announcement and the interview. But the scale of the funding confirms that the next chapter of AI will not be limited to writing better answers. Bezos is betting that artificial intelligence can learn to design the physical world. For Puerto Rico, the question is whether it will be ready to use that technology without giving up control over the technical value that already exists in its industrial economy.