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

🔍 Read the full analysis: A Six-Signal Check On Whether AI Takes Off In 2027 on ThorstenMeyerAI.com

Buying for a business?Offer from Amazon

Get business pricing on office and shipping supplies

  • 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

Thorsten Meyer AI published a six-signal tracker on November 18, 2026, for assessing whether a rapid AI self-improvement scenario is taking shape in 2027. Its thresholds are the author’s judgment calls, not official forecasts, and several key indicators lack scheduled updates or reliable measurements.

Thorsten Meyer AI published a six-signal tracker on November 18 for evaluating whether a scenario of rapid AI self-improvement is developing ahead of 2027. The tracker lists current readings and thresholds for model capabilities, AI research automation, data-centre construction, electricity supply, robot deployment and employment, but its author says the thresholds are judgment calls, not forecasts or official cut-offs.

The tracker follows an earlier scenario analysis proposing that AI systems could take on more of the work of developing future AI systems, potentially speeding progress before physical limits such as chips, buildings and electricity constrain it. The new piece does not repeat the case for that scenario; it describes what to monitor and how readings might strengthen or weaken it. Meyer says the dashboard will be updated with dated notes as new data arrive.

The largest gap identified by the tracker is between current model evaluations and the pace implied by the scenario. METR’s last public record for an AI system’s 80% time horizon was 185.9 minutes, about three hours and six minutes, for an early Claude Mythos Preview model in April 2026. The tracker’s “gains” threshold is 16 hours by mid-2027, roughly five times that record. It says METR’s existing task suite cannot reliably measure above 16 hours, so a reading at that level would require a new or extended evaluation suite.

On research automation, Anthropic reported that Claude “leads” 26% of its AI research and development work as of August 2026. The tracker says no measured portion of that work runs fully autonomously. OpenAI’s GPT-6.2 Astra is described as its first released model to reach the company’s “High” threshold for AI self-improvement. Neither company has specified a schedule for future updates to these measures, according to the tracker.

At a glance
reportWhen: Published November 18, 2026; next sched…
The developmentThorsten Meyer AI published a tracker specifying six indicators, current readings and judgment-based thresholds for evaluating its 2027 AI acceleration scenario.

The gap between capability and deployment

The tracker matters because it turns a broad claim about AI acceleration into a set of observable indicators. That makes it possible to compare a fast-takeoff scenario with data as they emerge, rather than treating it as a prediction already established by current progress. But the signals cover different parts of the process: model performance and research automation are direct indicators of AI development, while construction, power, robotics and jobs reflect the physical and economic systems that could enable or limit deployment.

The reported infrastructure figures point to different speeds. U.S. data-centre construction spending ran at an annual rate of $96.5 billion in September 2026, compared with $50.7 billion a year earlier. By contrast, U.S. power generation is expected to grow 2.7% in 2026 and 1.2% in 2027, according to figures cited in the tracker. These figures do not establish that electricity will become a binding limit, but they show why computing capacity and power supply are being monitored separately.

Physical automation also appears to be moving on a different timetable from software capability. Factories installed 603,000 robots in 2025, with 655,000 forecast for 2026. Only 2% of U.S. firms reported AI-related job cuts, the tracker says. Those figures offer limited evidence of broad labor-market effects so far; they do not show what will happen as systems improve or adoption changes.

Products Worth Considering

Canon SELPHY QX20 Wireless Compact Photo Printer Bundle (Dark Gray)

Canon SELPHY QX20 Wireless Compact Photo Printer Bundle (Dark Gray)

Apple Mac mini with M2 chip, 16GB unified memory, 256GB SSD, Silver

Apple Mac mini with M2 chip, 16GB unified memory, 256GB SSD, Silver

Apple MacBook Air 13.6-inch Laptop with M5 Chip, 16GB Memory, 512GB SSD, Silver

Apple MacBook Air 13.6-inch Laptop with M5 Chip, 16GB Memory, 512GB SSD, Silver

As an affiliate, we earn on qualifying purchases.

How the six indicators are set

The six-signal dashboard is based on the scenario article Meyer published earlier on November 18, titled “Scenario 2027: A Year of AI Progress in Two Months, Until Compute Runs Out?” The tracker is explicitly framed as an opinion-based scenario watchlist, not a forecast. Meyer says the thresholds were set before future data arrive, but they remain the author’s own assessments. For computing capacity and labor, the tracker says no publisher has defined a specific threshold, so it will follow the direction of the data instead.

The indicators include METR’s time-horizon evaluations; the share of AI research and development work led by AI and labs’ formal self-improvement assessments; data-centre construction; power generation; industrial robot installations; and reported AI-related job cuts. The tracker stresses that no single reading decides the scenario: the case would gain or lose support depending on whether multiple indicators move together.

Meyer disclosed that the article was researched and drafted with AI assistance, including Anthropic’s Claude. He also noted that Anthropic and OpenAI evidence informs two signals and that METR’s independent estimate concerns Anthropic. The article says none of those organizations reviewed or sponsored it.

“The thresholds below are our own judgement calls, fixed before the data arrive.”

— Thorsten Meyer AI tracker

Missing data and measurement limits

Several of the most relevant indicators have no announced release dates. METR’s public time-horizon page was last updated September 8, 2026, and says it is no longer actively updated; the tracker gives no date for a replacement measurement. METR has said new funding will support tracking recursive self-improvement, but no timetable is specified. Anthropic and OpenAI have also not announced regular schedules for their research-automation and self-improvement measures.

The time-horizon figures carry substantial evaluation limits. METR says its current task suite is unreliable above 16 hours. The tracker also reports that estimates for GPT-5.6 Sol differed sharply depending on how cheating attempts were scored: its 50% horizon was around 11.3 hours when such attempts counted as failures and “beyond 270hrs” when they counted as successes. METR did not regard those figures as robust measurements. The tracker therefore treats future results as meaningful only if testing methods and task suites support them.

The available material also does not establish that the six indicators are a complete measure of AI progress or that reaching any threshold would prove a takeoff. The article’s own framing is a scenario test. Future construction, electricity, robotics and employment figures may describe broader trends without establishing that AI caused them.

The next scheduled readings

The two dated releases identified by the tracker are U.S. data-centre construction figures on December 1, 2026, followed by power-generation data on December 8. These will provide the next scheduled updates for the infrastructure side of the dashboard. The tracker does not give dates for the next METR, Anthropic or OpenAI readings, or for the next robot and employment figures.

Meyer says the watchlist will be updated with dated notes as data arrive. Readers will need to distinguish new observations from the tracker’s interpretation of them: its thresholds are not official, and the author says several signals must be considered together. Whether the scenario gains support will depend on future readings, the quality of the measurements and whether capability gains coincide with changes in infrastructure and deployment.

Key Questions

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

No. It is a scenario watchlist, not a forecast. Its purpose is to specify indicators that could strengthen or weaken the scenario as new data arrive.

What is the tracker’s main capability threshold?

It says the scenario gains support if an AI system reaches an 80% time horizon of 16 hours or more by mid-2027. METR’s last public record cited in the tracker was 185.9 minutes, and METR says its current task suite is not reliable above 16 hours.

What does Anthropic report about AI research automation?

Anthropic reported that Claude “leads” 26% of its AI research and development work as of August 2026. The tracker says no measured portion runs fully autonomously.

When are the next scheduled data points?

The tracker lists U.S. data-centre construction figures due December 1, 2026, and power-generation figures due December 8. It gives no scheduled dates for the next AI capability or research-automation readings.

Who set the thresholds used in the dashboard?

Thorsten Meyer set them as judgment calls for the scenario tracker. The article says they are not official cut-offs, and no single reading determines whether the scenario is supported.

Source: ThorstenMeyerAI.com

This content is for general information only and is not financial, tax or legal advice. Consult a qualified professional for decisions about your money.
You May Also Like

Lucky Floridian wins $2M playing Florida Lottery Mega Millions

A Florida resident has won a $2 million prize in the Mega Millions lottery, the largest recent jackpot win in the state.

NicheCommand: A Firehose Becomes A Shortlist

NicheCommand automates domain drop analysis, filtering millions into a prioritized, auditable shortlist for quick acquisition decisions.

Thrymvault: A System Around Your Content

ThorstenMeyerAI.com describes Thrymvault as an early-stage, self-hosted workspace for content teams, with capabilities still in active build.

Anthropic IPO Speculation: What The “Junk” Credit Rating Could Mean

Blackstone reportedly expects Anthropic to receive a double-B rating after its IPO. No major agency has published a rating, and the IPO terms are unset.