Each bar runs from 0 to 25 years. Grey: recent median gap for research begun since 2000. Teal: plausible range with AI and digital twins. These are scenario assumptions, not forecasts.
Google DeepMind’s GNoME predicted 2.2 million new crystal structures (2023). Microsoft and PNNL used AI to screen 32 million candidates and make a new battery electrolyte (2024). AI-designed drug candidates are progressing through clinical trials. AI-generated layouts are used in Google’s TPU chips. Network digital twins are entering planning and connections work.
Factories, clinical trials, safety cases, standards, spectrum, planning consent and grid connections. That is why physical fields shrink less than software, and why regulation and assurance increasingly set the pace.
Creates a three-slide PowerPoint report from the current filters: the headline, a field-by-field comparison and the AI outlook. Charts are native and editable. Works offline.
Tracks chosen from this computer cannot be saved in a view; use a file name in the repo for that.
The sources consulted for this model. Recent milestones were checked against the web sources below during development. Historical dates draw on standard references; where a date is less certain, its details card says “Dates approximate”. Links open in a new tab.
Also: NCSC, Timelines for migration to post-quantum cryptography (2025); NIST FIPS 203–205 (2024); NCSC Cyber Assessment Framework.
Also: IBM quantum roadmap (2026), treated as a target rather than a forecast.
Other net worth figures are recent rich-list estimates; founding and listing dates come from company histories.
Nobel Prize citations (physics, chemistry, medicine); IEEE milestones and Engineering and Technology History; institutional histories of Bell Labs, IBM Research, Xerox PARC, CERN, DARPA, CSIRO, NPL, RAE Farnborough, the MRC Laboratory of Molecular Biology and UK Atomic Energy Authority; company histories and annual reports; regulator approval records (FDA, MHRA).
Select a single field (use “Only”) to see and filter its sub-fields.
Pick topics to show only those; with none picked, everything shows.
Diamonds mark arcs that have helped, or could help, deliver the September 2026 vision. The vision window breaks them down by area.
Everything in the dataset that is not on the display: retired, disputed, partly confirmed, waiting for QA, rejected, and possible duplicates.
Personal project. Built by Stuart Okin in a personal capacity; not affiliated with or endorsed by any employer. AI-maintained data, from public sources only. The dataset is compiled and kept up to date by AI agents using Claude (Anthropic), without human review, so it may contain errors. Outlook years are AI estimates, not forecasts. Check anything important against the cited sources.
Each arc traces one breakthrough from the first published or demonstrated research to its first commercial product, operational use or binding change. Lanes group arcs by where the research began (universities, industry labs, or government and publicly funded labs) or by field. The height of each arc is the gap in years. “Today” is the date the data was last updated, and the outlook always runs ten years beyond it.
Opens a panel you can move, resize and keep open.
Stuart Okin. A personal project, not affiliated with or endorsed by any employer.
Untick to hide a single arc. Select a name to inspect it.
Select an arc or a milestone to see its story, or search above.
Every change the data agents have made, newest first. Click a milestone to open its details. Highlighting in the model and during Play is set in Display.
Each arc runs from first research to first product or change. Taller means slower.
Dotted arcs are expected to land in the next ten years.
Fading arcs have not landed yet.
History: median years from research to change for arcs landing in the previous ten years. Beyond today: a scenario for research starting now, at the recent pace and with AI and digital twins. Follows your filters.
HistoryRecent paceWith AI and digital twins
Years from founding to the first listing, sale or private valuation that made founders multi-millionaires on paper. Net worth is the lead founder’s approximate wealth in 2026 money (Forbes 2026 where available; deceased founders at death, inflation-adjusted). Hover a figure for its basis. Follows the field filter.
DESNZ’s Vision for an AI-enabled clean energy system (call for evidence, September 2026, closing 6 November) groups AI opportunities into four areas and highlights barriers such as data access, incentives, regulatory clarity, security and skills. Select an area to show its arcs; select a name to inspect it.
Could not load QDayIntro.mp4. Put it in the same GitHub folder as index.html, or choose the file from this computer.
Could not load cae_cover_16x9.mp4. Put it in the same GitHub folder as index.html, or choose the file from this computer.
This device or browser can’t show the 3D model. Try a recent version of Chrome, Edge, Safari or Firefox on a laptop or desktop.
How long does research take to change the world, and is AI making it faster?
AI-maintained from publicly available sources, without human review, so it may contain errors.
Where would you like to start?
Narrated walk-throughs of one to two minutes, with subtitles. Each returns you here when it ends.
Each question opens the tool set up with the answer: a chart, its insight and the items behind it.