How To Tell If 2027 Will Be The Year AI Takes Off: Six Signals
AIThis post was created with the assistance of artificial intelligence (AI).

TL;DR

Thorsten Meyer AI published a November 18, 2026 watchlist tracking six indicators that could support or weaken its 2027 AI acceleration scenario. Current readings show rapid data-centre construction but slower power growth, limited reported AI-led research work, and little evidence of AI-related job cuts; the author’s thresholds are judgments, not forecasts.

Thorsten Meyer AI published a six-signal tracker on November 18 for assessing whether its proposed 2027 AI acceleration scenario is taking shape. The watchlist covers model capabilities, AI use in AI research, independent estimates of progress, data-centre construction, industrial robots and reported job cuts; its thresholds are the author’s judgments, not a forecast.

The tracker follows a separate scenario article that argues AI systems could increasingly help build the next generation of AI systems. That article describes a possible feedback loop in which progress that once took a year could arrive in about two months, followed by slower gains as physical limits such as chips, buildings and power constrain development. The tracker does not make that case again; it names measurements, sources and thresholds that Meyer AI says would strengthen or weaken it.

Before you orderOffer from Amazon

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
Start your free Prime trial Free trial for eligible customers · Cancel anytime
As an affiliate, we earn on qualifying purchases.

The current readings vary. METR’s last public record for an 80% task-completion time horizon was 185.9 minutes, for an early Claude Mythos Preview model in April 2026. Anthropic said Claude “leads” 26% of its AI research and development work in August, with no measured portion fully autonomous. METR’s October estimate of independent improvement on Anthropic’s benchmark was around 2x, with a stated range of 1.3x to 3x.

Infrastructure data show a different pace from capability measures. US data-centre construction spending reached a $96.5 billion annual rate in September 2026, compared with $50.7 billion a year earlier. The source says US power generation is expected to grow 2.7% in 2026 and 1.2% in 2027. Separately, the International Federation of Robotics counted 603,000 industrial robot installations in 2025 and forecast 655,000 for 2026. US Census Bureau data cited in the tracker found 2% of firms reported AI-related job cuts in the November 2025–January 2026 period.

At a glance
analysisWhen: Published November 18, 2026; figures an…
The developmentThorsten Meyer AI published a six-signal tracker on November 18, 2026, to monitor whether its proposed 2027 AI acceleration scenario is gaining support.
Log-scale chart titled The gap: about five times today's record. METR's measured 80% time horizons of record-setting AI models rise from 0.89 minutes for GPT-4 in March 2023 to 185.9 minutes for an early Claude Mythos Preview in April 2026, the last public record. A dashed line marks 16 hours; a highlighted target marks our gains line of 16 hours by mid-2027. A dotted line marks today, November 18, 2026, with no newer public record
Signal 1: METR's 80% time horizon records and our gains line. Source: METR, Time Horizon 1.1 results file and time horizons page (last updated September 8, 2026). Chart: Thorsten Meyer AI.

The Gap Between Software and Infrastructure

The six indicators test whether progress is spreading beyond model benchmarks into research workflows and the physical systems needed to run AI. A sharp rise in task horizons or AI-led research could support the scenario’s claim of a faster improvement loop. Construction growth alone would not establish that: power availability is a separate constraint, and the cited forecasts show slower growth in generation than the September construction annual rate might suggest.

The employment measure offers another check on how quickly AI is changing work across firms. The reported 2% figure is a specific survey observation, not a measure of all workplace automation or future job effects. Taken together, the readings can help readers distinguish signs of technical progress from evidence that deployment is changing infrastructure, factories or employment.

De’Longhi All-in-One Espresso and Coffee Maker with 10-Cup Pot and Milk Frother

De’Longhi All-in-One Espresso and Coffee Maker with 10-Cup Pot and Milk Frother

As an affiliate, we earn on qualifying purchases.

How the Six Measures Were Set

The tracker follows Meyer AI’s article, “Scenario 2027: A Year of AI Progress in Two Months, Until Compute Runs Out?” The source labels the new piece a scenario tracker and says the thresholds were set as the author’s judgment calls before the relevant data arrive. It also describes the scenario as neither a forecast nor investment advice.

The scenario’s central comparison is demanding: a year’s progress compressed into roughly two months implies about six times the prior pace. Against that, METR’s estimate for improvement on Anthropic’s tasks was around twice the earlier rate, and METR said it would not be surprised by a figure from 1.3x to 3x. The source also notes that METR’s task suite cannot reliably measure horizons beyond 16 hours, the tracker’s mid-2027 threshold.

The data do not arrive on one schedule. Meyer AI lists December 1, 2026, for the next construction-spending release and December 8 for the next power figures. It gives no announced schedule for new METR or lab readings. The tracker’s sources include METR, Anthropic, OpenAI, the US Census Bureau and the International Federation of Robotics.

““This piece does not argue the case again. It is the tracker.””

— Thorsten Meyer AI, describing the tracker

What the Current Readings Cannot Show

The indicators do not establish that the proposed acceleration will happen. The thresholds are Meyer AI’s own judgments, and the measures differ in scope and timing. METR’s public task suite has a stated measurement limit, its page was described as no longer actively updated, and new AI-specific results have no published release schedule in the supplied material.

Several details also remain open: whether labs will report comparable measures of AI-led research, whether task horizons will rise to the tracker’s 16-hour line, and whether power forecasts will change as data-centre construction expands. The cited job-cut figure covers a defined survey period and does not show what will happen to employment in 2027. The source states that Anthropic, OpenAI and METR did not review or sponsor the article, which was researched and drafted with AI assistance.

The December Infrastructure Releases

The next scheduled readings identified by Meyer AI are the US construction-spending release on December 1, 2026, and power figures on December 8. The tracker says it will add dated notes as data arrive. It does not provide dates for the next METR estimate or for updates from Anthropic and OpenAI, so those signals will depend on when the organizations publish new information.

Readers can compare incoming results with the tracker’s stated lines: a measured 80% time horizon of at least 16 hours by mid-2027; AI research work led well beyond Anthropic’s reported 26%, including fully autonomous work; and independent acceleration well above 3x. For infrastructure, the tracker looks for continued construction growth alongside raised power forecasts, while its robot and employment signals call for results clearly above their current cited levels. Those are the author’s scenario criteria, not confirmed predictions.

Key Questions

What is the six-signal tracker about?

It monitors six indicators that Thorsten Meyer AI says could help assess its 2027 AI acceleration scenario: task time horizons, AI involvement in AI research, independent estimates of acceleration, data-centre construction, robot installations and AI-related job cuts.

Does the tracker predict that AI will take off in 2027?

No. The source calls it a scenario tracker, not a forecast. Its thresholds are the author’s judgment calls for deciding whether the scenario is gaining or losing support.

What are the latest capability and research readings in the source?

METR’s last public 80% task time horizon was 185.9 minutes for an early Claude Mythos Preview model in April 2026. Anthropic said Claude led 26% of its AI research and development work in August, with none of the measured work fully autonomous.

When are the next readings due?

The tracker lists December 1, 2026, for US construction-spending data and December 8 for power figures. It gives no announced release schedule for updated METR or lab readings.

Source: Thorsten Meyer AI

You May Also Like

The Future of AI in Medical Imaging: Revolutionizing Healthcare

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

AI Management Skills Under Scrutiny: What a Coding Leaderboard Can’t Tell You

Real-world AI deployment demands management skills—reading internal data, maintaining honesty, and handling crises—beyond what traditional scoring benchmarks reveal.

Mastering the Art of Protecting Your Data in the Age of AI

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

The Ethical And Practical Reasons To Use The Best AI Model Instead Of Focusing On Sovereignty

A July 16 analysis argues most organizations should favor capable AI models and simple failover over costly sovereign infrastructure.