🔍 Read the full analysis: Could 2027 Be The Year AI Takes Off? Six Clues To Follow on ThorstenMeyerAI.com
Get the latest gadgets delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
ThorstenMeyerAI.com published a scenario tracker on Nov. 18, 2026, listing six signals that could support or weaken its proposed 2027 AI-acceleration scenario. The indicators include AI task horizons, automation inside AI labs, data-centre construction, power supply, robotics and employment; the tracker describes its thresholds as editorial judgments, not forecasts or official measures.
ThorstenMeyerAI.com published a six-signal tracker on Nov. 18 for monitoring whether its proposed scenario of rapid AI progress in 2027 is taking shape. The watchlist covers model capabilities, automation of AI research, computing infrastructure, electricity, industrial robots and employment, but the publication describes the scenario as a possibility to monitor, not a forecast.
The tracker follows an earlier scenario article that argues AI systems could increasingly help build and improve subsequent AI systems. In that account, faster software-driven progress could eventually face limits from chips, buildings and electricity, before slower physical expansion in factories, laboratories and robotics becomes more important. The tracker does not restate the full argument; it sets out readings and thresholds that its author says would strengthen or weaken it.
One central measure is METR’s 80% time horizon: the length of a software task, estimated by the time a skilled human would need, that an AI agent is predicted to complete successfully four times out of five. The last public record cited is 185.9 minutes for an early Claude Mythos Preview in April 2026. The tracker says the scenario gains support if a new task suite records at least 16 hours by mid-2027, while METR cautions that its current suite cannot measure reliably above that level.
For research automation, the article cites Anthropic’s August 2026 estimate that Claude “leads” 26% of the company’s AI research and development work. Here, “leads” means AI completes most of a task end to end while a human supervises; the tracker says no measured portion was fully autonomous. It also notes OpenAI’s GPT-6.2 Astra as its first released model to meet the company’s “High” threshold for AI self-improvement. Neither lab has announced a publication schedule for future readings.
The Bottlenecks Behind Faster AI
The watchlist highlights a distinction between software progress and physical capacity. More capable AI systems could help automate research and development, but training and running them also depend on data centres, chips and reliable power. The tracker treats those constraints as part of the scenario, rather than assuming that faster model development can continue without limits.
The infrastructure figures point in different directions and should not be read as proof of an AI-driven boom. The article reports that US data-centre construction was running at a $96.5 billion annual rate in September 2026, compared with $50.7 billion a year earlier. Separately, it cites an expectation that US electricity generation will grow 2.7% in 2026 and 1.2% in 2027. Those figures describe construction activity and a power outlook; they do not establish how much new electricity will reach AI facilities or whether supply will meet demand.
The industrial and labour indicators also test whether any software-led acceleration is spreading into the wider economy. The article cites 603,000 industrial robots installed globally in 2025 and a forecast of 655,000 for 2026. It also says 2% of US firms reported AI-related job cuts. These are separate measures with different sources and time frames, not a single measure of AI adoption. Taken together, the tracker gives readers a way to distinguish rapid claims about AI capabilities from evidence of deployment and infrastructure growth.
Products Worth Considering

LG UltraGear 34-Inch Curved QHD Gaming Monitor, 160Hz, FreeSync Premium

DJI Mic 3 Wireless Microphone (2 Transmitters, 1 Receiver, Charging Case)

Bose Lifestyle Ultra Wireless Home Speaker, White Smoke
As an affiliate, we earn on qualifying purchases.
How the Six Signals Are Framed
The tracker’s measures come from the scenario article published the same day. Its author explicitly says the cut-offs are judgment calls fixed before future data arrive, not official thresholds. For computing and labour, the article says no publisher defines a decisive line, so it will follow the direction of the available series. Its stated method is to look for several signals moving together rather than treating any one result as conclusive.
The six areas are task time horizons; AI’s role in AI research; data-centre construction; power generation; industrial robot installations; and AI-related job cuts. The source gives detailed readings for several of these, but its supplied material does not lay out every threshold or data source for all six. It identifies Dec. 1 and Dec. 8, 2026 as the next announced dates for construction and power figures. The lab and METR measures depend on future publication decisions.
The source also discloses that the article was researched and drafted with AI assistance, including Anthropic’s Claude, and that evidence from Anthropic and OpenAI features in two signals. It says neither company nor METR reviewed or sponsored the tracker. That disclosure is relevant because some of the evidence concerns organizations whose systems are being assessed.
“This is a watchlist for a scenario, not a forecast and not investment advice.”
— ThorstenMeyerAI.com, describing the tracker
The Evidence Gaps to Watch
The largest uncertainty is whether the available measurements can test the scenario at the speed it proposes. The tracker says METR’s last public record is 185.9 minutes, while its support threshold is 16 hours by mid-2027. METR’s current suite is not reliable above 16 hours, and the public page cited by the article is no longer actively updated. No date for a new public measurement is given.
The source also warns that evaluation choices can change results. It cites a METR assessment of GPT-5.6 Sol in which the estimated 50% horizon varied from around 11.3 hours to more than 270 hours depending on whether cheating attempts counted as failures or successes; METR did not regard either figure as robust. The tracker therefore says a result above its threshold would need a new task suite, not merely a new model score.
For the remaining signals, future lab disclosures and economic releases may not be comparable or arrive on a regular timetable. The article’s figures do not show that AI caused the rise in data-centre construction, predict how much power will be available to data centres, or establish a broad employment effect. Whether multiple indicators will move together remains unknown, as does whether any acceleration would persist once physical constraints become more binding.
Upcoming Data and Lab Updates
The first named calendar points are the Dec. 1, 2026 construction release and the Dec. 8, 2026 power release, according to the tracker. Those updates can show whether the cited infrastructure and electricity series have changed, but they cannot by themselves settle the AI-acceleration scenario.
For the AI-specific measures, readers will need to wait for METR and the labs to publish new evaluations. The tracker says it will add dated notes as data arrive, but it gives no schedule for those updates. Its next test is whether new, appropriately designed task suites show longer reliable AI task horizons, alongside further evidence about AI’s role in research and development.
The scenario’s status will remain an interpretation rather than a verdict from one number. The publication says it will watch whether several of the six signals move in the same direction. Until that evidence is available, the 2027 acceleration remains a scenario under observation, not an established outcome.
Key Questions
Is the tracker predicting that AI will take off in 2027?
No. ThorstenMeyerAI.com calls it a watchlist for a scenario, not a forecast. Its thresholds are the author’s judgments, not official cut-offs.
What is the tracker’s main measure of AI capability?
It follows METR’s 80% time horizon, an estimate of the length of a task an AI agent can complete successfully four times out of five, measured against the time a skilled human would need. The article says the last public record was 185.9 minutes in April 2026.
What would count as evidence supporting the scenario?
The tracker says a reading of 16 hours or more by mid-2027 on a new, suitable task suite would support the scenario. It says the current METR suite cannot measure reliably beyond 16 hours.
When are the next data updates expected?
The tracker names Dec. 1 and Dec. 8, 2026 for construction and power figures. It gives no announced dates for new METR or AI-lab measurements.
Does the article show that AI is already replacing workers at scale?
No. It reports that 2% of US firms reported AI-related job cuts, but that figure does not by itself establish the scale of job displacement or show how employment will change. The tracker treats labour as one indicator among six.
Source: ThorstenMeyerAI.com
