Know how fast AI is improving, and what comes next.
Model releases, capability results and benchmark scores as reported, next to the compute, power and money coming online. Then Spezia's outlook: where this is heading over the next 3 to 24 months, with a confidence on each call.
Free on Explorer. No card.
From the current outlook
Real output from Spezia, read from the same data the tool uses. Every line links to its source.
AI's binding constraint has shifted from model capability to physical and political delivery: capital commitments are locked in for years while newly elected US policymakers and local voters can slow the power, permits and chips those commitments depend on.
- Frontier capabilities6 monthslow confidence
AI-driven research acceleration will remain near the METR estimate of about 2x (1.97, range 1.3-2.8) rather than compounding sharply, because the time-horizon test is saturating (doubling time 129 days).
Basis: METR's 5 November review reports a 1.97x acceleration factor and a saturating time-horizon test (source 5).
- Compute, chips and power12 monthsmedium confidence
Gas turbine supply will stay the hardest power bottleneck through 2027: projected orders of 129 GW against 60-70 GW of global production capacity will keep pushing buyers to second-hand turbines and converted jet engines.
Basis: Source 15 reports 129 GW of projected turbine orders versus 60-70 GW of production capacity, with buyers already scouring for used turbines.
- Money and adoption12 monthslow confidence
Enterprise AI deployment spending will continue its steep climb past the $177 billion reported for the past year, with frontier labs holding about 69% of deployment spend.
Basis: Source 19 reports spending surging from $33B to $177B in one year with 69% frontier-lab share.
Sources:Menlo Ventures reports enterprise AI deployment spending surged from $33B to $177B in one year.
Projections, not facts. Each states its basis and horizon, and its confidence is the lower of two independent checks.
What you get
What shipped
The strongest model releases and capability results of the last 30 days, one line per story, each linked to its source.
Benchmarks, as reported
The best score each model has been reported at, with the quote it came from. Labs report their own numbers, and Spezia shows you where each one came from.
Where it is heading
Projections with a horizon and a confidence, each checked against its sources before it is shown.
How it works
- 1Read the outlook's one-line thesis.
- 2Scan the releases and benchmark results behind it.
- 3Ask Spezia what a result means for your product, team or portfolio.
Who uses it
Questions about AI Progress
Where do the benchmark numbers come from?
From the labs' own announcements and from independent reports, each shown with the words it was read from. Spezia does not run the benchmarks itself.
How is confidence decided?
The model that writes a projection gives one confidence and an independent check gives another. Spezia shows the lower of the two.
Do I need to pay?
No. AI Progress is part of Explorer, the free plan. Create an account with your name, email and a password and you land in it straight away.
Research from this data
Analysis written from what AI Progress tracks, with the sources and the day each figure was read.
- AI research and securityAnalysis1.9xMETR's estimate of how much faster Anthropic's benchmark scores rose
Is AI speeding up AI research? What METR found at Anthropic, and what it did not
METR's November 2026 review of Anthropic: benchmark scores growing 1.9 times faster, what that does and does not measure, its limits, and METR's time horizon.
Often used with
What to build, where to deploy AI and where capital is going next.
See it liveAI Intel
FreeTen areas of AI, read in full and checked, one line per story.
See it liveTooling guide
NewFreeThe best model and tool for each kind of work, rewritten weekly.
See it live
Know how fast AI is improving, and what comes next.
Free on Explorer, with 5 questions a month to Ask Spezia. No card.