EL7.AI
StocksMedium•19 November 2026•
6 min read

Anthropic’s $1.5 Trillion Valuation Tests the Market’s AI Boom

Key Facts

1Anthropic priced its offering at $83 a share, valuing the company at about $1.5 trillion, and raised $81 billion.
2OpenAI’s announced improvement to AMD’s leading GPU design reached as much as 14%.
3METR estimated that growth in Anthropic model benchmark scores accelerated 1.9 times since October 2025, with a 95% confidence interval between 1.5 and 2.8 times.
4Claude’s share of Anthropic’s AI research tasks rose from less than 1% in February 2026 to more than a quarter in August.
5Anthropic’s commitments for computing and infrastructure spending totaled $900 billion in the coming years, including $70 billion over the following twelve months.

Anthropic priced its offering at $83 a share, valuing the company at about $1.5 trillion on a fully diluted basis in an offering that raised $81 billion, Reuters reported. Trading was expected to begin on Thursday, November 19, 2026, on Nasdaq under the ticker “ANT,” according to the Associated Press. The size and value of the offering put the AI company at the forefront of a test of investor appetite for assets tied to the sector. The pricing coincides with announcements of faster use of models in AI research and chip design. OpenAI had announced that its “GPT-6.2 Astra” model had reached a high threshold in its ability to help improve AI and accelerate research. These developments bring company valuations into focus alongside the costs and capabilities behind them.

OpenAI said “GPT-6.2 Astra” was the first model it had released externally to reach a “high” level on its AI self-improvement metric, which measures a model’s ability to accelerate research in the field. The company said the rating was equivalent, in its assessment, to making an additional highly capable research engineer available to each of its researchers, compared with the 2024 baseline. This provides a practical measure of the impact the company expects the model to have within its research teams. In a separate announcement about its collaboration with AMD, OpenAI said improvements to the company’s leading GPU had reached as much as 14% compared with already-optimized baselines. It added that these gains represented about a year of progress in less than five months. According to OpenAI, AMD incorporated some changes into a design sent for fabrication and made other changes to production design files for upcoming versions.

Anthropic’s indicators offer another picture of the acceleration of model-supported research. METR estimated that the company’s model benchmark scores have risen since October 2025 at 1.9 times the previous rate, with a 95% confidence interval ranging from 1.5 to 2.8 times. The organization said this rate was roughly equivalent to two years of historical progress in one year. The estimate is based on a model-capabilities index that the organization said is the version of the Epoch capabilities index used by Anthropic. The estimate does not mean that every aspect of research advanced at the same pace; it measures benchmark scores over the period defined by the organization. These findings provide context for the market’s bet on Anthropic, while keeping the estimate of the pace of progress tied to its metric and duration.

METR said it believed most of the observed acceleration was likely due to Anthropic’s intensive use of AI in research and development, but added that there was significant uncertainty about this interpretation. It said human assistance remained the primary approach, rather than fully independent contributions from models. In Anthropic data cited by Axios, Claude’s share of the company’s AI research tasks rose from less than 1% in February 2026 to more than a quarter of tasks in August. The data indicated that humans continued to oversee these tasks. This shift signals broader use of the model within the company itself, without indicating that research has moved to fully autonomous work. Axios also reported that Google’s Gemini is reducing chip-planning work that used to take weeks to hours, pointing to the expansion of model use from software research to hardware design.

On model comparisons, OpenAI said “GPT-6.2 Astra” scored 5.2 points higher than “GPT-6 Astra” on the “Last Exam for Agents” test, at less than a third of the cost. It added that the model scored 5.5 points higher than “Claude Fable 5.2” at just 5% of its cost on the test. These figures link performance to test cost, a comparison the company presented when it launched the model. The results are not directly related to Anthropic’s offering price, but come amid broader competition over model capabilities and operating costs. The pairing of model advances with chip-design applications means companies are presenting gains in research and engineering as part of AI’s value to the economy. The comparison figures remain attributed to the organization that announced them.

The AI Futures Project forecast that machines would surpass the best human AI researchers next year, Axios reported. Paul Christiano, who recently joined OpenAI’s board, also estimated a one-in-three chance of superintelligence emerging within a year if development continued at its current pace. These are forecasts, not outcomes that have already occurred. On potential obstacles, METR said it had seen no evidence of major bottlenecks in algorithmic progress, training compute, experiments or data over the next twelve months. It added, however, that its understanding of the obstacles had changed more than once as additional information came in, and that it was not confident in its current assessment. This brings together estimates of accelerating progress with caveats about how well future constraints can be identified.

The scale of financial commitments adds another dimension to Anthropic’s valuation. The company had taken on $6.2 billion in debt and announced $900 billion in commitments for computing and infrastructure spending in the coming years, according to the Associated Press. That includes $70 billion over the following twelve months, the report said. These commitments put the cost of the infrastructure needed to develop and operate models alongside the offering’s value of about $1.5 trillion. In a comparison cited by the agency, SpaceX’s offering raised about $75 billion in June 2026, while its market value was about $1.9 trillion on Wednesday afternoon, November 18. By this comparison, Anthropic’s offering exceeded the amount raised in SpaceX’s offering, while the market value cited for SpaceX remained higher.

In market reaction on November 19, U.S. futures rose after the offering was priced; Nasdaq futures climbed 1.3% and S&P futures rose 0.36%, Reuters reported. In Asia, Taipei’s benchmark index fell 4.4%, and the Asia-Pacific index excluding Japan hit its lowest level in seven months. U.S. futures and Asian stocks moved in opposite directions in the session after the pricing. Anthropic shares were expected to begin trading on Nasdaq on Thursday, November 19, under the ticker “ANT,” according to the announcement of the pricing. The Associated Press said the availability of shares would allow employees to begin selling vested stock, but described the company’s market performance as the biggest test after pricing. Trading will provide the first direct market measure of investor acceptance of the offering’s valuation.

Open on chart