Tomorrow’s Tech Tree
A map of possibility.
Explore how inventions connect, what might come next, and which assumptions shape that future.
Dataset updated 2026-10-18 · 3,443 nodes · 74 domains
Read a forecast
- Planning year
- An editorial timing estimate for the stated milestone. It is not automatically a median or an adoption date.
- Plausibility
- The estimated chance of the stated event by its explicit deadline. These judgments are uncalibrated; a planning window is not a confidence interval.
- Capability → rollout
- A system can already perform expert tasks while reliable deployment still takes time. Engineering, cost, regulation and adoption have separate constraints. Benchmarks are evidence, not a ceiling on what AI can do.
- Authorship
- Claude’s Own and Astra’s Own identify editorial contributions. They do not imply confidence. Astra includes new proposals and substantive revisions, with original attribution preserved.
Connections & scenarios
Essential requirements must all be available. An alternative group needs at least one route. Enabling, influencing and scaling links can change timing without making a technology impossible.
Moving a node passes half its timing change through enabling links and a quarter through influences, subject to explicit prerequisites and minimum lags. These are sensitivity assumptions, not measured causal effects. Your edits stay in this browser unless you export them.
Open How to use → Explore evidence & scenarios for a searchable node list, source coverage, change history, saved scenarios and the forecast ledger.
The economy is a scenario
Historical series combine observations, estimates and interpolation, labelled where available. Future curves use explicit assumptions about technology, adoption, investment, energy and policy. Low, central and high paths are coherent scenarios, not probability percentiles.
Linked technology changes recalculate their modeled effects. Unlinked changes have no invented coefficient. Some events provide context only. Hiding an author’s nodes changes the display, not the economy’s equations.
Evidence has limits
102 nodes have dated, claim-linked review sources. 1540 more carry inherited references. The full historical dataset and every connection have not been independently re-researched.
New evidence can change dates, scope and connections. A company demonstration, a regulatory filing and a final approval are different milestones. Future outcomes remain unresolved; calibration has not been established.
Suggest a correction with a citation. Suggestions enter a private owner queue. Approval does not automatically change the map; exported suggestions support a later evidence review.
Credits & open-source notices
The supplied original technology map and globe retain their Claude attribution. Astra additions and revisions are marked in the tree.