Jensen Huang says Nvidia achieved AGI, again — not that it matters
Nvidia chief executive Jensen Huang said on the company’s earnings call that it had “achieved AGI” for many tasks, but immediately described such milestones as “senseless”. The article argues that this reflects a wider problem: artificial general intelligence has no agreed definition or measurable threshold, allowing companies to make ambitious claims that may primarily serve to promote AI progress.
Huang has made similar remarks before, though he later acknowledged that AI agents could not independently build a company like Nvidia. OpenAI defines AGI broadly as autonomous systems outperforming humans at most economically valuable work, while a separate reported agreement with Microsoft uses a $100 billion profit benchmark; other firms use terms such as “superintelligence” and “useful general intelligence”. Huang instead emphasised AI’s practical and commercial value, including productive work, token generation and the demand for more computing power.
- Huang says Nvidia has achieved AGI for some tasks.
- AGI remains undefined and difficult to measure.
- Companies increasingly frame AI progress in commercial terms.
New here? Start with this
Artificial general intelligence, or AGI, is a term for AI that can handle a wide range of tasks at a level comparable to or beyond people. There is no shared definition or agreed test for when a system has reached it, so claims about AGI can mean different things.
Jensen Huang leads Nvidia, which makes computer chips widely used to develop and run AI systems. Other technology companies have set out their own descriptions or measures of AGI, while Nvidia often discusses AI in terms of tasks it can perform and its commercial uses.
The distinction matters because claims about AGI can shape public expectations about how capable AI is and how quickly it is advancing. They can also affect how companies explain the value of AI and the computing resources needed to run it.
Both sides, in good faith
The strongest fair case each way — we don't pick a winner.
The case for
Supporters of Huang’s position argue that AGI labels are less important than whether AI systems reliably deliver useful economic and scientific work. If models can perform a broad and growing range of valuable tasks at or above human level, treating the achievement as meaningful in practice can be reasonable even without a single universally accepted threshold. They also contend that focusing on deployment, productivity and the computing needed to support it is more concrete and honest than pursuing a potentially arbitrary semantic finish line.
The case against
Critics argue that saying a company has achieved AGI is misleading when neither the term nor a test for it has a settled definition, especially if the same systems cannot independently carry out complex open-ended work such as building a company. Ambiguous claims can blur the distinction between impressive task performance and genuinely general autonomy, making it harder for investors, policymakers and the public to assess capabilities and risks. On this view, companies should describe specific, independently verifiable abilities rather than use a powerful label whose meaning can shift to suit commercial incentives.