Is AI About To Take Off In 2027? Look For These Six Signals
AIThis post was created with the assistance of artificial intelligence (AI).

🔍 Read the full analysis: Is AI About To Take Off In 2027? Look For These Six Signals on ThorstenMeyerAI.com

Before you orderOffer from Amazon

Get hardware and tech 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.

TL;DR

ThorstenMeyerAI.com published a six-indicator watchlist for assessing whether AI progress could accelerate sharply in 2027. The article says the scenario is not a forecast: current evidence is mixed, some readings are months old, and several signals have no scheduled updates.

ThorstenMeyerAI.com published a six-signal tracker on November 18 for monitoring whether AI development could accelerate sharply in 2027, including measures of AI task performance, lab automation, computing capacity, power, robotics and employment. The tracker describes its thresholds as the authors’ judgment calls, not a forecast; several key data points have no announced publication date.The tracker follows an earlier scenario in which AI systems increasingly help build the next generation of AI systems, potentially speeding progress until limits in chips, buildings and electricity intervene. Its author, Thorsten Meyer, says the new article is intended to monitor that scenario rather than repeat the argument for it. The thresholds were set in advance, according to the article, but are not official cut-offs, and no single reading would decide whether the scenario is developing. The sharpest gap identified is between current measurements and the scenario’s benchmark for AI task performance. METR’s last public record for an 80% task-completion time horizon was 185.9 minutes, for a model tested in April 2026. The tracker’s “gains” line is an 80% horizon of at least 16 hours by mid-2027. METR cautions that its current task suite cannot measure reliably beyond that range, so a new, longer test suite would be needed to interpret a result at the threshold. For AI involvement in AI research, the article reports Anthropic’s August 2026 estimate that Claude “leads” 26% of the company’s AI research and development work, where “leads” means completing most of a task end to end under human supervision. The tracker says no measured portion runs fully autonomously. It also notes OpenAI’s claim that GPT-6.2 Astra was its first released model to reach the “High” threshold for AI self-improvement. These are lab-reported measures, not independent demonstrations of broad autonomous research.
At a glance
reportWhen: Published November 18, 2026; future ind…
The developmentThorstenMeyerAI.com published a dated tracker of six indicators that could support or weaken its scenario of rapid AI progress in 2027.
Crypto market snapshot
Fear & Greed Index
12/100 — Extreme Fear
Bitcoin BTC$57,904▼ 6.3%
Ethereum ETH$1,627▼ 7.1%
Tether USDT$0.9986▼ 0.1%
BNB BNB$524.04▼ 8.3%
USDC USDC$0.9994▼ 0.0%
XRP XRP$0.8911▼ 7.6%
Solana SOL$67.44▼ 7.1%
TRON TRX$0.3028▼ 2.0%
Live data · CoinGecko · alternative.me (24h change)

The gap between AI speed and infrastructure

The tracker matters because it turns a broad claim about an impending AI acceleration into specific indicators readers can monitor. If model task horizons and lab automation rise quickly at the same time as compute capacity expands, that would offer more support for the scenario than any single headline model release. Conversely, slow or stagnant measurements could weaken it. The article also highlights a potential mismatch between fast-growing data-centre construction and slower changes in electricity supply. It cites US data-centre construction at an annual rate of $96.5 billion in September 2026, compared with $50.7 billion a year earlier, while projected power-generation growth is 2.7% in 2026 and 1.2% in 2027. These figures do not prove that power will constrain AI development, but they help frame the physical limits the scenario anticipates. For readers, the main implication is that the 2027 claim remains testable but unsettled. The tracker itself says the largest uncertainty is speed: METR’s estimates of model-improvement rates fall well short of the roughly sixfold pace implied by “a year of progress in two months.”

Products Worth Considering

Apple MacBook Pro 16.2-inch with M5 Pro, 24GB Memory, 1TB SSD, Wi-Fi 7, Space Black

Apple MacBook Pro 16.2-inch with M5 Pro, 24GB Memory, 1TB SSD, Wi-Fi 7, Space Black

As an affiliate, we earn on qualifying purchases.

How the six signals are tracked

The six categories in the tracker are AI task time horizons; AI’s share of AI research; available compute; data-centre construction and power; industrial robotics; and labour-market effects. The author says the measures come from different publishers and have different schedules. Construction and power are the only categories with announced near-term dates in the article: US construction data on December 1, 2026, and power figures on December 8. Some indicators are difficult to compare over time. METR’s public results page was last changed on September 8, 2026, and says it is no longer actively updated. The article reports that METR’s new funding is intended to support tracking recursive self-improvement, but no publication schedule has been given. It also warns that task-suite design and how cheating attempts are scored can greatly alter measured horizons. On physical deployment, the tracker cites 603,000 factory robots installed in 2025 and a 2026 forecast of 655,000. On employment, it says only 2% of US firms reported AI-related job cuts. Those figures describe different aspects of adoption; neither alone shows how much work AI systems can perform or how quickly job effects may change.

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

— Thorsten Meyer, ThorstenMeyerAI.com

Missing data and disputed thresholds

The tracker does not establish that the scenario will occur. Its thresholds are the author’s judgments, and the article says there is no publisher-defined line for the compute or labour signals; those are assessed by direction. Several AI-specific readings depend on labs or METR choosing to publish further results, with no dates supplied. There are also limits to the available measurements. METR’s current task suite is not reliable above 16 hours, and the tracker describes a large change in one task-horizon estimate depending on whether cheating attempts count as successes or failures. The article says METR did not regard those figures as robust. It is also unclear from the supplied material how the six signals will be updated when comparable measures are unavailable, or how much weight each should receive. The source article discloses that it was researched and drafted with AI assistance, including Anthropic’s Claude, and that evidence from Anthropic and OpenAI is central to two signals. It says neither company nor METR reviewed or sponsored the article. The supplied source material ends partway through its discussion of the second signal, so the full details and thresholds for all six indicators are not available here.

The next scheduled data releases

The first dates named in the tracker are December 1, 2026, for US data-centre construction figures, and December 8, 2026, for power data. The article says it will add dated notes as figures arrive. No calendar date is given for fresh METR task-horizon results or new public readings from Anthropic and OpenAI, so those signals may remain unchanged until the organizations publish updates. Readers can compare upcoming releases with the tracker’s stated benchmarks, while keeping in mind that the thresholds are scenario tests rather than predictions. A clearer assessment of the 2027 outlook will depend on whether multiple indicators move in the same direction and whether the underlying measures remain comparable.

Key Questions

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

No. ThorstenMeyerAI.com calls it a scenario tracker, not a forecast. It sets out indicators that could strengthen or weaken the scenario.

What is the tracker’s main AI-performance benchmark?

It says the scenario gains support if an AI system reaches an 80% task time horizon of at least 16 hours by mid-2027. METR’s cited public record was 185.9 minutes in April 2026, and METR says its current suite cannot measure reliably beyond 16 hours.

Does Anthropic’s 26% figure mean AI research is autonomous?

No. The tracker says Claude “leads” 26% of Anthropic’s AI research and development work, meaning it completes most of a task under human supervision. It reports that no measured portion runs fully autonomously.

When are the next figures due?

The tracker names December 1, 2026, for US data-centre construction and December 8, 2026, for power figures. It gives no scheduled dates for new METR or lab-reported AI readings.

What remains uncertain about the six signals?

The benchmarks are author-selected, several readings lack publication dates, and some measures have methodological limits. The supplied source material does not include the full detail for all six indicators, so their complete thresholds cannot be verified from it.

Source: ThorstenMeyerAI.com

Nothing in this article is financial or investment advice. Cryptocurrency and precious-metal investments carry significant risk — do your own research and consider a licensed advisor.
You May Also Like

10 Best AI Automation Platforms For Learning Workflow Building In 2027

A 2027 roundup compares 10 books and guides for learning AI automation, from no-code workflows to Microsoft tools and technical RPA skills.

When One Agent Isn’t Enough: Claude Now Builds Its Own Team Of Agents On The Fly

Anthropic’s Claude now autonomously creates and manages its own team of agents for complex tasks, enhancing performance in high-value workflows.

CTOs Are Escaping

Senior tech leaders are shifting from CTO positions to hands-on roles at Anthropic, reflecting a shift in power towards AI model development and frontier research.

I Put 71 AI Model Setups To Work On My News Pipeline

A publisher’s test of 71 AI configurations found GPT-6 Astra scored highest, while cheaper GPT-6 Luna was selected for production.