TL;DR
Thorsten Meyer AI published a tracker on November 18, 2026, for six indicators that could test its 2027 AI acceleration scenario. The current readings show rapid data-center construction but a large gap between measured AI progress and the scenario’s most ambitious thresholds; several indicators have no scheduled update.
The tracker follows six indicators: how long AI systems can perform tasks at an 80% success rate; AI’s role in AI research; an independent estimate of model improvement speed; data-center construction; industrial robot installations; and reported AI-related job cuts. The publisher describes the tracker as a watchlist for a scenario, not a forecast. Its thresholds are the publisher’s own judgment calls, set before the data arrive.
Get everyday essentials delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
METR’s last public record for the 80% task horizon was 185.9 minutes, for an early Claude Mythos Preview model in April 2026. The scenario’s stated gain threshold is 16 hours by mid-2027. METR said its current task suite cannot reliably measure beyond that point, and its page is no longer actively updated, according to the tracker.
Anthropic reported in August that Claude “leads” 26% of its AI R&D work, with none of the measured work fully autonomous. OpenAI said on November 17 that GPT-6.2 Astra was its first released model to reach the company’s “High” threshold for AI self-improvement. The tracker also cites METR’s October estimate of roughly 2x acceleration on Anthropic, with a range of 1.3x to 3x. It says the scenario’s “a year of progress in two months” pace would imply about 6x.

Where the Indicators Point
The readings describe a mixed picture rather than a single confirmed trend. Data-center construction is rising quickly, while the tracker says US power generation is forecast to grow 2.7% in 2026 and 1.2% in 2027. Those figures matter because the scenario depends on computing capacity, buildings and electricity, while the article argues that physical constraints could limit faster progress.
The AI capability indicators also sit below the scenario’s most ambitious markers: METR’s roughly 2x estimate is below the tracker’s 6x illustration, and the public task-horizon record is far short of 16 hours. These measures do not establish what will happen in 2027. They give readers specific benchmarks against which future claims can be compared.
Employment and robotics offer a slower-moving check on whether AI progress is translating into wider economic change. The tracker cites 603,000 industrial robots installed in 2025, with 655,000 forecast for 2026, and says 2% of US firms reported AI-related job cuts. These are distinct measures: robot installations do not directly measure AI capability, and reported job cuts do not capture every effect of AI on work.
Products Worth Considering

DJI Mini 4K Drone with 4K UHD Camera

Logitech MX Keys S Combo Wireless Keyboard and Mouse, Graphite

LG 27GS60QC-B Ultragear 27-inch Curved Gaming Monitor (QHD, 180Hz, 1ms, 1000R)
As an affiliate, we earn on qualifying purchases.
The Scenario Behind the Watchlist
The tracker follows an earlier Thorsten Meyer AI analysis titled “Scenario 2027: A Year of AI Progress in Two Months, Until Compute Runs Out?” That article proposed a possible feedback loop in which AI systems do more work on building the next generation of AI systems. It also described a later slowdown if limits on chips, buildings and power become binding, followed by a longer physical buildout involving robots, factories and labs.
The November 18 article says it is not repeating the case for that scenario. Instead, it lists what to measure, the sources, the readings available at publication and the conditions under which the scenario would gain or lose support. The publisher disclosed that the article was researched and drafted with AI assistance, including Anthropic’s Claude, and that Anthropic, OpenAI and METR did not review or sponsor it.
Among the economic readings, the tracker reports a September 2026 US data-center construction annual rate of $96.5 billion, compared with $50.7 billion a year earlier. It attributes the construction figure to the US Census Bureau. These are annual-rate figures for construction spending, not a count of facilities.
““A scenario is only useful if you can tell early whether it is coming true.””
— Thorsten Meyer AI
Limits in the Current Readings
The tracker does not establish that the proposed acceleration will occur. METR’s estimate concerns Anthropic and is a range; it is not a measure of every AI lab or all AI work. The task-horizon record is also limited by the current suite’s stated measurement range, and METR’s page is no longer actively updated.
Several readings depend on companies or researchers choosing to publish new information. The tracker gives no fixed schedule for updates on AI research automation or independent acceleration. Its thresholds, including 16 hours and the conditions for a signal to “gain,” are the publisher’s own choices rather than standards set by METR, Anthropic or OpenAI.
The available source material does not provide the complete tracker table or a full schedule for every signal. It identifies December 1, 2026, for the next construction figures and December 8 for the next power figures; it does not state the precise release details in the supplied text. The 2% job-cuts figure is attributed to a Census Bureau survey covering November 2025 to January 2026, so it is not a current measure of layoffs.
December Releases and Later Updates
The tracker lists December 1, 2026, for the next data-center construction figures and December 8 for power figures. It does not give set dates for fresh METR, Anthropic or OpenAI readings, or for the next robot-installation forecast.
Thorsten Meyer AI says it plans to update the tracker with dated notes as data arrive. Those updates may show whether the indicators move toward its stated thresholds, remain below them or point in different directions. Until then, the six readings are a monitoring framework for a proposed scenario, not confirmation of what 2027 will bring.
Key Questions
What did Thorsten Meyer AI publish?
It published a six-indicator tracker on November 18, 2026, to monitor a proposed scenario of faster AI progress in 2027.
Does the tracker predict that AI progress will accelerate in 2027?
No. The publisher calls it a watchlist for a scenario, not a forecast. Its thresholds are its own judgment calls.
What is the latest public AI task-horizon reading cited?
The tracker cites METR’s 185.9-minute 80% time horizon for an early Claude Mythos Preview model from April 2026. It says METR’s current task suite cannot reliably measure the scenario’s 16-hour threshold.
When are the next scheduled figures due?
The tracker lists US data-center construction figures for December 1, 2026, and power figures for December 8. It gives no fixed schedule for the AI-specific updates.
Source: Thorsten Meyer AI
