Six Trends That Could Indicate Whether AI Takes Off In 2027
AIThis post was created with the assistance of artificial intelligence (AI).

🔍 Read the full analysis: Six Trends That Could Indicate Whether AI Takes Off In 2027 on ThorstenMeyerAI.com

Buying for a business?Offer from Amazon

Get business pricing on tech for your team

  • Business-only prices and quantity discounts
  • Tax-exempt purchasing
  • Multiple users, one account, clear invoices
As an affiliate, we earn on qualifying purchases.

TL;DR

ThorstenMeyerAI.com published a six-signal tracker on November 18, 2026, for monitoring a scenario in which AI helps build more capable AI systems and progress accelerates in 2027. The tracker says current evidence falls short of its most demanding speed threshold, and stresses that its thresholds are the author’s judgments, not official forecasts.

ThorstenMeyerAI.com published a six-signal tracker on November 18 for judging whether a scenario of sharply accelerated AI progress is taking shape in 2027. The tracker compares current public evidence with thresholds set by its author, Thorsten Meyer, and says the biggest gap is between current AI task performance and the scenario’s proposed pace; Meyer describes the exercise as a watchlist, not a forecast.

The tracker follows six areas: AI agents’ performance on longer tasks; AI’s role in research and development at AI labs; formal assessments of models’ self-improvement capabilities; data-centre construction; electricity supply; and the spread of robots and AI-related job cuts. Meyer says no single indicator would decide whether the scenario is unfolding. The case would strengthen if several signals moved in the predicted direction.

For AI task performance, the last public record cited is a 185.9-minute 80% time horizon for an early Claude Mythos Preview model in April 2026. The scenario’s “gains” threshold is a 16-hour horizon by mid-2027, roughly five times that record. The article notes that METR, which measures AI agents, says its current task suite cannot reliably measure results above 16 hours. Its public results page was last updated September 8, and no next release date is listed.

Anthropic reported that, as of August 2026, Claude “leads” 26% of the company’s AI research and development work. In the tracker, “leads” means AI completes most of a task end to end while a person supervises; it does not mean the work is fully autonomous. The source also says OpenAI’s GPT-6.2 Astra is its first released model to reach the company’s “High” threshold for AI self-improvement. These are lab assessments, not independent measurements of AI development across the industry.

At a glance
reportWhen: Published November 18, 2026; several ke…
The developmentThorstenMeyerAI.com published a dated watchlist of six indicators intended to track whether a rapid-AI-progress scenario for 2027 is developing.

The gap between AI speed and infrastructure

The tracker’s central question is whether gains in AI capability can compound quickly enough to change the pace of research and development. If AI systems take on more of the work involved in producing later systems, the scenario argues, progress could accelerate. But the article’s own figures show why that remains uncertain: the proposed 16-hour task-performance threshold is far above the last public record, and the measurement tools may need to change before such a result can be judged reliably.

Physical capacity is another constraint. The source reports that US data-centre construction reached a $96.5 billion annual rate in September 2026, compared with $50.7 billion a year earlier. It cites expected US electricity-generation growth of 2.7% in 2026 and 1.2% in 2027. Those measures point to a mismatch the scenario says could matter: computing facilities can be built quickly, while power supply may expand more slowly. They do not, by themselves, establish how much computing capacity will be available to AI developers.

The labor and robotics measures provide a check on claims that AI is already changing the wider economy at the same pace. The source says factories installed 603,000 robots in 2025, with 655,000 forecast for 2026, and that 2% of US firms reported AI-related job cuts. These are distinct indicators with different coverage; neither proves that AI-driven automation is accelerating across all sectors.

Products Worth Considering

Bose QuietComfort Wireless Bluetooth Noise-Cancelling Headphones, Ice Blue

Bose QuietComfort Wireless Bluetooth Noise-Cancelling Headphones, Ice Blue

Logitech MX Keys S Combo Wireless Keyboard and Mouse, Graphite

Logitech MX Keys S Combo Wireless Keyboard and Mouse, Graphite

As an affiliate, we earn on qualifying purchases.

How the six-signal tracker works

The tracker accompanies an earlier scenario article by Meyer, “Scenario 2027: A Year of AI Progress in Two Months, Until Compute Runs Out?” That article proposes a chain in which AI systems increasingly contribute to building subsequent AI systems, until limits in chips, buildings and power become binding. It then envisages a slower phase tied to physical systems such as robots, factories and laboratories. The tracker is designed to test that scenario against future data, rather than restate the argument for it.

Meyer says the thresholds are his own judgment calls, fixed before the data arrive, and not official cut-offs. For computing capacity and labor, the article says no publisher defines a threshold, so it proposes watching the direction of the data instead. The source identifies December 1, 2026, for the next construction figure and December 8 for the next power-generation figure. Updates from METR and the AI labs have no announced schedule.

The source also discloses that the article was researched and drafted with AI assistance, including Anthropic’s Claude. It says evidence from Anthropic and OpenAI is central to two signals and that METR’s independent estimate concerns Anthropic. The source states that Anthropic, OpenAI and METR did not review or sponsor the tracker.

“This is a watchlist for a scenario, not a forecast and not investment advice.”

— Thorsten Meyer, in the November 18 tracker

Evidence gaps in the 2027 scenario

The supplied source material ends partway through its account of the second signal, so it does not provide full details for all six indicators, including every current reading, threshold and publication schedule. The tracker’s headline summary names the areas it follows, but that is not enough to reconstruct the complete dashboard or independently verify each underlying data series.

There is also no confirmed basis here for treating the scenario as a prediction. The 16-hour line is a judgment call, and METR’s stated limits mean a new task suite would be needed to assess results at or above that level reliably. The source flags the potential for cheating to distort task-horizon estimates, and says METR did not consider the cited GPT-5.6 Sol measurements robust.

Lab metrics have additional limits: “leads” describes supervised work, not fully autonomous research, and the tracker reports no measured portion of Anthropic’s work running fully autonomously. The source does not establish that the reported lab results represent AI developers generally. It also does not show how much of the construction and electricity trends will be available to AI computing specifically.

Next releases and dated checks

The next scheduled public data points identified in the source are US data-centre construction figures on December 1, 2026, and electricity-generation figures on December 8, 2026. Meyer says he plans to update the tracker with dated notes as new figures arrive.

For the AI-specific indicators, readers will have to wait for METR and the labs to publish updates; the source gives no dates. A meaningful test of the scenario’s task-performance line would require a longer task suite that can measure results beyond the current stated limit. The tracker’s stated approach is to judge the scenario from multiple signals together, rather than treating any single release as decisive.

Key Questions

Is the tracker forecasting that AI will take off in 2027?

No. Meyer calls it a watchlist for a scenario, not a forecast. Its thresholds are the author’s judgments rather than official cut-offs.

What is the tracker’s main test for faster AI progress?

One main test is whether AI agents can complete much longer software tasks reliably. The source cites a last public 80% time-horizon record of 185.9 minutes and sets 16 hours by mid-2027 as the scenario’s gains threshold.

Does Anthropic’s 26% figure mean its AI research is autonomous?

No. The source says Claude “leads” 26% of Anthropic’s AI research and development work, meaning it completes most of a task while a human supervises. It says no measured portion runs fully autonomously.

When are the next scheduled data releases?

The tracker identifies US data-centre construction figures due December 1, 2026, and power-generation figures due December 8. It lists no announced dates for the next METR or AI-lab updates.

Source: ThorstenMeyerAI.com

You May Also Like

How Consumer AI Hardware Could Split Into Utility and Prestige

AIThis post was created with the assistance of artificial intelligence (AI).Consumer AI…

200 Hong Kong Services to Adopt AI by 2027

Uncover how Hong Kong’s plan to implement AI in over 200 services by 2027 could transform your daily life and future opportunities—continue reading to learn more.

ByteDance’s New Scientist Program: Who Are They Eyeing In AI?

ByteDance launches a scientist program aimed at recruiting elite young AI researchers, amid competition with OpenAI for top mathematical talent.

Epoch AI Showed How AI Can Turn Firefox Fixes Into Working Attacks

Epoch AI says GLM 5.3 built attacks against six Firefox releases using public fixes. Mozilla patched a separate flaw in Firefox 159.