THE YOTTABIT SCALE INDEX / THINK 10²⁴
Numbers that make
you say “WOW.”
What happens when AI becomes hundreds of times cheaper? When biology becomes searchable? When millions of robots are already at work? The figures are real research snapshots. The opportunities are what we make of them.
THE UNIMAGINABLE IS ALREADY HAPPENING
Seventeen ways
the future got bigger.
Read the headline. Feel the scale. Then explore the source, the limits and why it might matter to your work and your life.
AI economics and capability01 / 10²⁴
286×
CHEAPER AI AT THE SAME BENCHMARK
From $20 to seven cents.
A comparable level of AI capability became dramatically less expensive in under two years.
Genomics and molecular biology02 / 10²⁴
~190,000×
LESS EXPENSIVE TO SEQUENCE
The cost of reading a human genome collapsed.
The production cost fell from roughly $95 million in 2001 to about $500 by 2022.
Genomics and molecular biology03 / 10²⁴
200M+
AI-PREDICTED STRUCTURES
Two hundred million protein shapes to explore.
Researchers can search a vast database of protein structure predictions.
Automated scientific discovery04 / 10²⁴
2.2M
COMPUTATIONAL CANDIDATES
An AI system imagined 2.2 million crystal structures.
A model proposed millions of possible inorganic crystals for scientists to investigate.
Robotics and autonomy05 / 10²⁴
542,000
NEW INSTALLATIONS
More than half a million factory robots arrived in a year.
Industrial robots were installed worldwide in 2024—more than twice the annual number a decade earlier.
Robotics and autonomy06 / 10²⁴
4.66M
OPERATING INDUSTRIAL ROBOTS
There are millions of robots already at work.
The global operational stock of industrial robots reached roughly 4.66 million in 2024.
Energy and storage07 / 10²⁴
93%
LOWER INSTALLED COST
Grid-scale batteries became dramatically cheaper.
The average fully installed utility-scale battery project cost fell from about $2,571 to $192 per kWh between 2010 and 2024.
Energy and storage08 / 10²⁴
950 TWh
2030 IEA SCENARIO
The digital world may need twice as much data-center electricity.
The IEA projects roughly 950 TWh in 2030, compared with an estimated 485 TWh in 2025.
Deployment and diffusion09 / 10²⁴
20M+
ELECTRIC CARS SOLD
Over twenty million electric cars sold in one year.
Global electric car sales exceeded twenty million vehicles in 2025.
Information, networks and sensing10 / 10²⁴
6B
PEOPLE ONLINE
Six billion people were connected to the internet.
The ITU estimated approximately six billion internet users in 2025.
Information, networks and sensing11 / 10²⁴
2.2B
NOT YET CONNECTED
And 2.2 billion people were still offline.
Even with record global connectivity, roughly 2.2 billion people remained offline in 2025.
Economic, social and environmental12 / 10²⁴
122,511
AI PATENT GRANTS IN 2023
AI-related patents grew from thousands to over 120,000.
Stanford’s 2025 AI Index reports 122,511 AI patents granted in 2023, versus 3,833 in 2010.
AI economics and capability13 / 10²⁴
4–5×
COMPUTE GROWTH PER YEAR
The biggest AI training runs multiplied rapidly.
A historical analysis found frontier model training compute rose roughly four-to-fivefold annually across a selected sample through May 2024.
AI economics and capability14 / 10²⁴
88%
SURVEYED ORGANIZATIONS
Nearly nine in ten surveyed organizations use some AI.
In a 2025 survey, 88% of participating organizations reported using AI in at least one function.
Automated scientific discovery15 / 10²⁴
80,150
AI-SUPPORTED NATURAL SCIENCE PAPERS
Tens of thousands of science papers now involve AI.
The 2026 AI Index reports about 80,150 AI-supported publications in natural sciences during 2025.
Energy and storage16 / 10²⁴
169 GWh
NEW STORAGE CAPACITY
A battery-storage boom measured in gigawatt-hours.
The research register reports 169 GWh of grid-scale battery additions commissioned globally in 2024.
Automated scientific discovery17 / 10²⁴
38.8%
A SCIENCE BENCHMARK
AI science agents still have a long way to go.
A leading science agent scored 38.8% on a particular evaluation versus 83.5% for PhD experts, as summarized in the 2026 AI Index.
Why we show the sources: Some figures describe measured history, others estimates or projected scenarios. They are not all exponential, and we won't pretend a prediction has already happened. Each story links to its research context.
THE UNIVERSE OF SIGNALS
There's more
to discover.
We've found more extraordinary questions in the research. If you like digging into the evidence, the full source collection is here too.
Explore the evidence behind YottaBit OPEN THE 100-SIGNAL RESEARCH LIBRARY +
38source-linked research snapshots
62questions still being investigated
100signals being tracked across the Yottabit Era
Go deeper: the complete 100-signal research atlas
For readers who want the detail. Search, filter and follow individual sources. Preliminary signals are labeled as such—without dressing them up as discoveries.
38 signals shown
001The biggest AI training runs multiplied rapidly.Historical trend through May 2024 · Epoch AIEvidence available002From $20 to seven cents.Nov 2022–Oct 2024 · Stanford AI Index 2025Evidence available003Agent end-to-end task accuracyAn open question we are tracking, not yet a measured trendResearch in progress004Frontier model benchmark scoreAn open question we are tracking, not yet a measured trendResearch in progress005Agent task completion reliabilityAn open question we are tracking, not yet a measured trendResearch in progress006AI hardware price-performance30% annual cost decline in selected ML hardware sample; Stanford 2025Evidence available007AI hardware efficiency40% annual energy-efficiency gain in selected ML hardware sample; Stanford 2025Evidence available008Frontier training cluster power2.0x/year fitted historical power trend; 2024 studyEvidence available009Algorithmic efficiency at fixed benchmarkAn open question we are tracking, not yet a measured trendResearch in progress010Nearly nine in ten surveyed organizations use some AI.2025 survey · Stanford AI Index 2026Evidence available011The cost of reading a human genome collapsed.2001–2022 · NHGRI historical cost comparisonEvidence available012DNA sequencing cost per megabaseNHGRI longitudinal cost series beginning 2001Evidence available013Genome sequence production throughputAn open question we are tracking, not yet a measured trendResearch in progress014Number of reference genome assembliesAn open question we are tracking, not yet a measured trendResearch in progress015Population genomes sampledAn open question we are tracking, not yet a measured trendResearch in progress016Two hundred million protein shapes to explore.AlphaFold Protein Structure Database · consulted Oct 2026Evidence available017Protein structure experimental validation rateAn open question we are tracking, not yet a measured trendResearch in progress018Gene editing regulatory approvalsFirst US CRISPR-based therapy FDA approved Dec 8 2023Evidence available019Gene therapy clinical-trial enrolmentAn open question we are tracking, not yet a measured trendResearch in progress020Single-cell transcriptomic profiles depositedAn open question we are tracking, not yet a measured trendResearch in progress021Robotic lab experiments completed per day353 synthesis recipes across 17-day campaign; 2023 Nature paperEvidence available022Autonomous materials synthesis target successNature 2023 paper and subsequent discussion disagree on validationEvidence available023An AI system imagined 2.2 million crystal structures.GNoME · 2023Evidence available024Predicted stable candidate structures380,000 high-stability candidates in 2023 reportEvidence available025Independent experimental confirmations of predicted structuresAn open question we are tracking, not yet a measured trendResearch in progress026Tens of thousands of science papers now involve AI.2025 · Stanford AI Index 2026Evidence available027Scientific paper replication accuracy by agentsAn open question we are tracking, not yet a measured trendResearch in progress028Research cycle duration from hypothesis to validated resultAn open question we are tracking, not yet a measured trendResearch in progress029Robotic lab successful synthesis yieldAn open question we are tracking, not yet a measured trendResearch in progress030AI science agents still have a long way to go.2026 report · Stanford AI IndexEvidence available031More than half a million factory robots arrived in a year.2024 · International Federation of RoboticsEvidence available032There are millions of robots already at work.2024 · International Federation of RoboticsEvidence available033Industrial robot densityAn open question we are tracking, not yet a measured trendResearch in progress034Service robot annual installationsAn open question we are tracking, not yet a measured trendResearch in progress035Autonomous mobile robots in commercial deploymentAn open question we are tracking, not yet a measured trendResearch in progress036Autonomous vehicle rider-only milesAn open question we are tracking, not yet a measured trendResearch in progress037Commercial robotaxi paid tripsAn open question we are tracking, not yet a measured trendResearch in progress038Robot fleet utilization rateAn open question we are tracking, not yet a measured trendResearch in progress039Human intervention frequency in robotic deploymentsAn open question we are tracking, not yet a measured trendResearch in progress040Multi-robot task throughputAn open question we are tracking, not yet a measured trendResearch in progress041Leading edge foundry process revenue shareTSMC 3nm = 24% wafer revenue in 2025Evidence available042Advanced-node mass production milestoneTSMC 2nm high-volume manufacturing Q4 2025Evidence available043Chip energy per inference taskAn open question we are tracking, not yet a measured trendResearch in progress044AI accelerator installed computational capacityAn open question we are tracking, not yet a measured trendResearch in progress045GPU effective memory bandwidthAn open question we are tracking, not yet a measured trendResearch in progress046Advanced memory bit densityAn open question we are tracking, not yet a measured trendResearch in progress047Training data centre power densityAn open question we are tracking, not yet a measured trendResearch in progress048Chiplet package interconnect throughputAn open question we are tracking, not yet a measured trendResearch in progress049Data centre capital investmentGlobal data-centre investment about $0.5 trillion in 2024 per IEAEvidence available050High-performance compute energy efficiencyAn open question we are tracking, not yet a measured trendResearch in progress051Six billion people were connected to the internet.2025 · International Telecommunication UnionEvidence available052And 2.2 billion people were still offline.2025 · International Telecommunication UnionEvidence available053Fixed-broadband subscriptionsAn open question we are tracking, not yet a measured trendResearch in progress054Mobile-broadband subscriptionsAn open question we are tracking, not yet a measured trendResearch in progress0555G population coverageMore than 50% of global population covered, 2025Evidence available056International internet bandwidthAn open question we are tracking, not yet a measured trendResearch in progress057Annual fixed broadband trafficAn open question we are tracking, not yet a measured trendResearch in progress058Annual mobile broadband trafficAn open question we are tracking, not yet a measured trendResearch in progress059Satellite Earth-observation data availabilityAn open question we are tracking, not yet a measured trendResearch in progress060Percentage of sensing data retainedAn open question we are tracking, not yet a measured trendResearch in progress061The digital world may need twice as much data-center electricity.2025 estimate / 2030 scenario · IEAEvidence available062AI-focused data centre electricity demandAn open question we are tracking, not yet a measured trendResearch in progress063Global renewable power capacity additionsAn open question we are tracking, not yet a measured trendResearch in progress064Solar PV levelized generation costUSD 44/MWh global weighted average, 2025Evidence available065Onshore wind levelized generation costUSD 33/MWh global weighted average, 2025Evidence available066Grid-scale batteries became dramatically cheaper.2010–2024 · IRENAEvidence available067A battery-storage boom measured in gigawatt-hours.2024 · research register citing industry energy dataEvidence available068Grid connection queue lengthAn open question we are tracking, not yet a measured trendResearch in progress069Energy storage duration at commercial scaleAn open question we are tracking, not yet a measured trendResearch in progress070Solar cell record conversion efficiencyAn open question we are tracking, not yet a measured trendResearch in progress071Over twenty million electric cars sold in one year.2025 · International Energy AgencyEvidence available072Electric vehicle share of new salesAround 25% of all new cars sold in 2025Evidence available073Industrial electricity demand growthAn open question we are tracking, not yet a measured trendResearch in progress074Electricity demand growth from data centres~70 TWh growth in 2025 per IEA Global Energy Review 2026Evidence available075Real-world AI adoption in clinical workflowsAn open question we are tracking, not yet a measured trendResearch in progress076Share of AI use with human oversightAn open question we are tracking, not yet a measured trendResearch in progress077Business agent deployment by functionSingle-digit adoption in most functions; 2025Evidence available078Robots installed per manufacturing workerAn open question we are tracking, not yet a measured trendResearch in progress079AI adoption gap across firm sizesAn open question we are tracking, not yet a measured trendResearch in progress080Training and digital-skill gapAn open question we are tracking, not yet a measured trendResearch in progress081Validated AI-proposed new drug candidatesAn open question we are tracking, not yet a measured trendResearch in progress082Autonomous synthesis successes validated independentlyAn open question we are tracking, not yet a measured trendResearch in progress083Clinical gene editing patients treatedAn open question we are tracking, not yet a measured trendResearch in progress084AI-enabled grid flexibility dispatchedAn open question we are tracking, not yet a measured trendResearch in progress085Robot fleet autonomous task hoursAn open question we are tracking, not yet a measured trendResearch in progress086AI-assisted battery compound experimental winsAn open question we are tracking, not yet a measured trendResearch in progress087Genomic diagnosis turnaround timeAn open question we are tracking, not yet a measured trendResearch in progress088Materials screening-to-prototype cycle lengthAn open question we are tracking, not yet a measured trendResearch in progress089Industrial design iteration timeAn open question we are tracking, not yet a measured trendResearch in progress090Energy per unit of validated scientific discoveryAn open question we are tracking, not yet a measured trendResearch in progress091Global AI corporate investmentUSD 581.69 billion total corporate AI investment in 2025Evidence available092Public research AI expenditureAn open question we are tracking, not yet a measured trendResearch in progress093AI task productivity change in field trialsAn open question we are tracking, not yet a measured trendResearch in progress094Share of science output with AI component~5.8%-8.8% by field in 2025Evidence available095Electricity sector low-emission generation growth2025 clean generation increase exceeded total global net generation growthEvidence available096Avoided fuel expenditures from renewable deploymentEstimated USD 480 billion 2025Evidence available097Net job creation in automation adopting firmsAn open question we are tracking, not yet a measured trendResearch in progress098Technology capability diffusion lagAn open question we are tracking, not yet a measured trendResearch in progress099Broadband affordability burdenAn open question we are tracking, not yet a measured trendResearch in progress100AI-related patents grew from thousands to over 120,000.2010 vs 2023 · Stanford AI Index 2025Evidence available
No signals match those filters. Try another term or category.