Will 2026 be the year artificial general intelligence (AGI) finally takes over, or are we just puffing up another massive tech bubble? In a recent eye-opening interview, AI pioneer Andrew Ng cuts through the noise to reveal what is genuinely moving the needle in the tech landscape. For value investors trying to separate transformational business utilities from fleeting marketing jargon, this conversation is an absolute goldmine.

The former head of Google Brain, co-founder of Coursera, and a foundational architect of modern AI. Ng

Ng’s insights provide a much-needed reality check for anyone looking to allocate capital intelligently in the tech space. Rather than getting swept up in the race for sci-fi intelligence, let's explore the most important arguments from the episode that will actually impact the real economy.

The AGI Hype and the New "Turing Test"

There is a massive disconnect between what the public thinks AGI means and how tech companies are using the term. Ng notes that AGI is increasingly being redefined just so companies can claim they’ve reached it, pointing out that, "AGI has become a marketing term rather than something of precise technical meaning." When asked if we will achieve AGI by 2026, he is remarkably candid: "For any reasonable definition of AGI, I think the answer is no." Instead of moving the goalposts, Ng proposes a highly practical, economic "Turing AGI Test." For an AI to truly be considered AGI, a human judge should be able to design a multi-day onboarding experience where the AI functions exactly like a remote worker, completing economically valuable tasks as well as a skilled professional. We are still decades away from that reality.

The True 2026 Catalyst is "Agentic Workflows"

While scaling up massive models has driven the generative AI revolution so far, the real business value moving forward lies in what Ng coins "agentic AI." Smarter base models are incredible, but if you simply let an AI loose with a bunch of tools, it lacks the reliability needed for enterprise use cases. Instead of waiting for a single monolithic intelligence to solve everything, developers are building highly specific agentic workflows. By taking complex mental processes—like checking tariff compliance or reading legal documents—and coding them step-by-step into AI agents, businesses can create systems that run reliably 10,000 times in a row. As Ng puts it, "In 2026 and beyond there'll be a lot of exciting work to build AI agents or to build agentic workflows to do a ton of really valuable economically important work."

The Looming Threat of an AI Hype Bubble

For retail investors who adhere to value investing principles, identifying a bubble is just as important as identifying a trend. Ng expresses genuine concern that the constant over-promising of AGI timelines could trigger another "AI winter." If businesses and investors expect human-level intelligence in two years and it doesn't arrive, the resulting crash in confidence and funding could be devastating. He warns, "I think excessive hype that leads to disappointment that leads to you know collapse of the so-called bubble—that would not be good for the world and good for the field of AI." Diffusing this hype is essential for sustainable, long-term growth in the sector.

Job Automation is About Tasks, Not Total Wipeouts

The fear that AI will immediately cause mass unemployment is vastly overstated. Ng clarifies that only a very small fraction of jobs—such as call center workers, translators, and voice actors—face near 100% automation. For the vast majority of complex professions, AI might only be able to automate specific components. Ng references task-based analyses, noting that, "For a lot of jobs, AI can automate like 30 to 40% of someone's job and so you still need a human to do that 60 to 70%." The real threat isn't the AI itself, but the human competitor utilizing it. Ng reinforces the famous modern tech adage: AI won't replace you, but someone who uses AI will replace someone who doesn't.

Open-Source Moats and Avoiding an AI Oligopoly

When assessing the competitive landscape of AI companies, the battle between proprietary closed models and open-source models is critical. Ng is a strong advocate for open-source and open-weight models because they prevent a small handful of tech giants from becoming the sole gatekeepers of innovation. He compares it to the current mobile market duopoly of iOS and Android, hoping AI avoids that fate. Interestingly, the global dynamics are shifting to support this open ecosystem, with Ng observing that "a lot of the best open source open weight models are coming out of China." Ensuring a thriving open-source community is the best way to guarantee a rich, diverse landscape of AI applications rather than an oligopoly that strangles third-party innovation.

The most actionable takeaway for a value-oriented investor is to ignore the sensationalist predictions of AGI and focus entirely on practical adoption. The companies that will generate true economic value in the coming years aren't necessarily the ones promising human-level intelligence tomorrow; they are the businesses meticulously building reliable, agentic workflows that solve tedious, everyday problems today. Focus your attention on the quiet implementers, not the loud prognosticators.

For more of my insights on this topic, be sure to follow me.