Archive · 8 Sep 2026
Silicon Valley Keeps Moving the Goalposts on AGI Claims
Some technology leaders say artificial general intelligence, or AGI, has arrived.
AGI usually means a computer that can think and do many kinds of work like a person.
The problem is that people disagree about what should count as AGI.
Jensen Huang and OpenAI executives used different achievements to support their claims.
Critics say the definition keeps changing whenever a new product appears.
The article argues that calling something AGI can help companies raise huge amounts of money.
It also says companies such as NVIDIA benefit when demand for computer chips and data centers grows.
The writer does not say the technology is fake or unimportant.
Instead, the writer asks readers to separate real progress from financial hype.
NVIDIA CEO Jensen Huang declared AGI had arrived twice under different criteria.
OpenAI leaders similarly suggested GPT-6 Astra marked the beginning of the AGI era.
Critics argue AGI lacks a stable, independently verifiable definition and is being diluted.
The article links AGI claims to enormous valuations, fundraising, and demand for computing infrastructure.
It acknowledges major technical progress while warning that financial incentives may inflate expectations.
- Who
- NVIDIA CEO Jensen Huang, OpenAI leaders Sam Altman and Greg Brockman, and critics including Demis Hassabis, François Chollet, and Yann LeCun are discussed.
- What
- Technology leaders and companies are making competing claims that artificial general intelligence has arrived, while critics challenge the term’s shifting definition.
- When
- The article cites claims made late last year, in March, May, and September 7, along with other recent financing announcements.
- Where
- The discussion centers on Silicon Valley and the broader AI industry; the article also references an Ohio data-center campus and Gurugram as an analogy.
- Why
- The author argues that AGI claims can support fundraising, high company valuations, and continuing demand for computing infrastructure.
AGI Advocates and Industry Leaders
AGI Skeptics and Definition Critics
Whether AGI has arrived
AGI Advocates and Industry Leaders
Jensen Huang, Sam Altman, and Greg Brockman have said recent AI systems justify describing the industry as being in, or having reached, the AGI era.
AGI Skeptics and Definition Critics
Demis Hassabis, François Chollet, and Yann LeCun question whether current systems demonstrate the broad, general intelligence implied by AGI.
Meaning of AGI
AGI Advocates and Industry Leaders
The term can be applied to systems that automate economically valuable work, operate as agents, or achieve strong performance on demanding tasks, according to definitions discussed in the article.
AGI Skeptics and Definition Critics
Critics favor a stricter definition involving the full range of human cognitive abilities and argue that routine-task automation or benchmark success is not intelligence itself.
Role of financial incentives
AGI Advocates and Industry Leaders
Large investments and massive computing projects reflect confidence that AI will eventually produce substantial economic value.
AGI Skeptics and Definition Critics
The author argues that AGI language helps companies justify extraordinary valuations, fundraising, and infrastructure spending before future benefits are demonstrated.
- AGI claim
- Jensen Huang said AGI had arrived in March and again on September 7, according to the article.
- OpenAI system
- The article identifies OpenAI’s GPT-6 Astra as the system associated with the September announcement.
- Computing scale
- The article says Astra was trained on more than 100,000 NVIDIA Grace Blackwell GPUs, with 400,000 additional GPUs discussed.
- OpenAI financing
- OpenAI reportedly closed $122 billion in committed capital at an $852 billion valuation in March.
- Anthropic financing
- Anthropic reportedly raised $65 billion at a $965 billion valuation in May.
- NVIDIA investments
- The article says NVIDIA invested $30 billion in OpenAI and made additional commitments involving CoreWeave and an Ohio data-center campus.
- Core criticism
- Critics cited in the article argue that AGI should involve broad human-level cognitive ability, not merely routine-task automation or benchmark performance.