TL;DR
A November 18, 2026, watchlist identifies six measures for tracking whether AI progress accelerates in 2027, from model capabilities to data-center construction, power, robots and jobs. The article presents thresholds as the publisher’s judgments, not a forecast; current evidence shows activity but does not establish that a year of progress will arrive in two months.
Thorsten Meyer AI published a six-signal watchlist on November 18 to track whether its proposed scenario of sharply faster AI progress is taking shape in 2027. The tracker lists current readings, sources and publisher-set thresholds, while describing the scenario as a possibility to monitor rather than a forecast.
Get business pricing on tech for your team
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
The indicators cover AI performance on longer tasks, AI’s role in AI research, independent estimates of model-improvement speed, data-center construction, industrial robot installations and reported AI-related job cuts. The publisher says a year of progress in two months would imply roughly sixfold acceleration; it cites METR estimates of about 2x acceleration for Anthropic, with a stated range of 1.3x to 3x. METR’s estimate does not reach the scenario’s implied pace.
On the capability measure, the tracker cites a 185.9-minute 80% time horizon as METR’s last public record, for an early Claude Mythos Preview model in April 2026. It sets 16 hours by mid-2027 as a threshold supporting the scenario. Anthropic reported in August that Claude “leads” 26% of its AI research and development work, with no measured work fully autonomous. OpenAI said GPT-6.2 Astra, released November 17, reached its “High” threshold for AI self-improvement.
Physical infrastructure indicators show investment and constraints moving at different rates. US data-center construction spending was running at a $96.5 billion annual rate in September, compared with $50.7 billion a year earlier, according to the Census Bureau. US power generation was forecast to grow 2.7% in 2026 and 1.2% in 2027. The International Federation of Robotics reported 603,000 industrial robot installations in 2025 and forecast 655,000 for 2026. Separately, 2% of US firms reported AI-related job cuts in Census Bureau data spanning November 2025 to January 2026.

The Gap Between Models and Infrastructure
The tracker’s central question is whether improvements in AI capability can move faster than the physical systems needed to support them. A sharp rise in model performance could affect research and development first, but wider deployment also depends on data centers, electricity and equipment. The cited construction total is an annualized rate, while the power figures are forecasts; neither by itself establishes how much usable capacity will be available to AI projects.
The labor reading also puts limits on claims about near-term economic effects. The 2% figure measures firms reporting AI-related job cuts over a stated survey period. It is not a count of workers displaced, a measure of all AI-related workplace changes, or evidence that automation has already transformed employment broadly. The six indicators therefore describe different parts of the scenario and should not be treated as one synchronized measure.
Products Worth Considering

DJI Mini 4K Drone with 4K UHD Camera

Apple AirPods Pro

Bose SoundLink Plus Portable Bluetooth Speaker, Black
As an affiliate, we earn on qualifying purchases.
A Scenario With Measurable Tests
The watchlist accompanies a separate scenario article published earlier on November 18. That article considers the possibility that AI systems could take on more of the work involved in building future AI systems, accelerating progress until constraints such as chips, buildings and power become binding. It further suggests robotics and manufacturing could matter in a later phase. The tracker says it is not repeating that argument; its purpose is to follow evidence that could support or weaken it.
Its thresholds—including the 16-hour task horizon and the points at which construction or robot deployment would count as support—are the publisher’s own judgment calls, fixed before the relevant future data arrive. The article discloses that it was researched and drafted with AI assistance, including Anthropic’s Claude, and that evidence from Anthropic, OpenAI and METR features in the analysis. It says those organizations did not review or sponsor the article.
“A scenario is only useful if you can tell early whether it is coming true.”
— Thorsten Meyer AI, describing the watchlist
What the Current Readings Cannot Show
The available indicators do not confirm that AI progress is accelerating toward the proposed pace. METR’s task-horizon page was last updated September 8, 2026, and the tracker says it is no longer actively updated; the task suite cannot reliably measure beyond 16 hours. The timing of new readings from METR and the AI labs is not set out, so several of the most direct indicators have no announced release schedule.
The tracker also does not establish that the separate measurements are directly comparable or that its chosen thresholds will predict broader economic change. Construction spending does not reveal how much data-center capacity is completed or available for AI, and a firm-level job-cut share does not measure total employment effects. The source gives no fuller detail here on the methods behind every indicator or on the precise conditions that would count as the scenario losing support.
December Data and Lab Updates
The next scheduled readings identified by the tracker are US construction spending on December 1, 2026, and a US power update on December 8. The source does not specify which power release is due that day. METR and the AI labs have not stated a schedule for their next relevant updates, so changes in those signals will depend on when the organizations publish new results.
Thorsten Meyer AI says it plans to add dated notes as data arrive. The watchlist will gain support if the specified capability, automation and infrastructure measures cross its stated thresholds; readings that stall or remain far below them would weaken the case. Those conclusions remain dependent on future data and on the publisher’s chosen criteria.
Key Questions
Is the article predicting that AI will take off in 2027?
No. The publisher calls it a scenario tracker, not a forecast, and says its thresholds are judgment calls.
What are the six signals?
They are AI performance on longer tasks, AI’s share of AI research, independent estimates of acceleration, data-center construction, industrial robot deployment and firms reporting AI-related job cuts.
What is the current evidence on AI acceleration?
The tracker cites METR’s estimate of about 2x for Anthropic’s model-improvement rate, with a range of 1.3x to 3x. It says the proposed pace of a year of progress in two months would be about 6x.
When are the next scheduled updates?
The tracker lists US data-center construction spending for December 1, 2026, and a power update for December 8. It gives no scheduled dates for new METR or AI-lab readings.
Source: Thorsten Meyer AI
