Progress you can inspect
See what has improved, where the biggest gaps remain and which work still needs evidence.
At a glance
Useful changes
Explore the compute access journey, three project evidence files and community question pack. The dated site change log records released changes.
Largest gaps
Sourced coordinates: 6 of 234. Recorded lifecycle stage: 66 of 234. These are overlapping source records, not unique facilities or proof of operation.
Inspect missing fields. Worldwide completeness and useful visitor outcomes remain unestablished.
Next work
Keep consequential news current, deepen project locations and stages, and connect developments to the records and decisions they affect.
See recorded improvement work. Email signup status is separate from the reviewed news feed.
How current is the information?
Latest recorded news: 2026-11-12. Last scoped news review: 2026-10-20. A scoped review is not a check of the whole market. Site changes do not refresh these dates.
Snapshot: snapshot available. Generated 2026-11-14T10:35:31.368Z. A snapshot becomes stale after 48 hours. Loading this page does not refresh it.
Inspect news coverage and uncovered areas · Inspect source monitoring.
Latest operational evidence
These are the latest verified receipts included in this public feed, not a live heartbeat. An enabled schedule is not proof of completion. A stale or missing receipt leaves current operation unverified here.
| Work | Evidence state | Latest verified success |
|---|---|---|
| Maintenance checkpoints | stale checkpoint Last verified failure: 17 Oct 2026, 07:37 UTC | 30 Oct 2026, 17:34 UTC |
| Analytics report | stale checkpoint | 17 Oct 2026, 06:20 UTC |
| Regional source scout | not yet verified | Not yet verified |
| Independent source review | not yet verified | Not yet verified |
| Scheduled maintenance | stale scheduled pickup | 25 Oct 2026, 14:30 UTC |
Inspect checkpoint dates, failures and scope
This page is serving. Receipt dates and site-change dates are separate. Evidence becomes stale after 48 hours independently of a new build. Read the health response.
The maintenance evidence is a restricted, allowlisted readback of a stored completion receipt, with independent review, automatic production and a live check. Raw receipts stay private. Analytics component delivery is separately scoped below; it does not establish all-report delivery or scheduled execution. A Search Console API block in the collector is separate from native Search Console review and page-indexing status. Research returns still need their own receipt adapters.
Maintenance checkpoints
Evidence state: stale checkpoint.
- Latest verified success included in this feed
- 30 Oct 2026, 17:34 UTC
- Last verified failure
- 17 Oct 2026, 07:37 UTC
- Success evidence readback
- 30 Oct 2026, 17:35 UTC
- Planned next evidence check
- Not recorded (not recorded)
Receipt-verified release or failed-checkpoint evidence, not a live heartbeat, broad outage claim or proof that all work is complete.
Failure stage: live verification. Failure evidence readback: 17 Oct 2026, 07:38 UTC.
Checkpoint revision: 4dfa6f4d0a90 (full revision in the public snapshot).
Analytics report
Evidence state: stale checkpoint.
- Latest verified success included in this feed
- 17 Oct 2026, 06:20 UTC
- Last verified failure
- Not recorded. This does not mean no failures occurred.
- Success evidence readback
- 17 Oct 2026, 06:21 UTC
- Planned next evidence check
- 18 Oct 2026, 06:15 UTC (check due)
GA component only: a saved private report was validated and accepted. Search Console remains blocked in this collector; native Search Console reviews are separate; scheduled execution is unverified. This is not all-report delivery, qualified audience, growth or visitor benefit. Evidence age uses the report observation, not acceptance or build time.
Report observed: 17 Oct 2026, 05:59 UTC. Reporting window: 17 Sept 2026 to 14 Oct 2026, inclusive (Etc/GMT).
Checkpoint revision: 9eb54354b8e1 (full revision in the public snapshot).
Regional source scout
Evidence state: not yet verified.
- Latest verified success included in this feed
- Not yet verified
- Last verified failure
- Not recorded. This does not mean no failures occurred.
- Success evidence readback
- Not yet verified
- Planned next evidence check
- 18 Oct 2026, 06:15 UTC (check due)
Natural delivery has not been verified in this public feed. A planned check is not promised delivery.
Independent source review
Evidence state: not yet verified.
- Latest verified success included in this feed
- Not yet verified
- Last verified failure
- Not recorded. This does not mean no failures occurred.
- Success evidence readback
- Not yet verified
- Planned next evidence check
- 18 Oct 2026, 06:15 UTC (check due)
Natural delivery has not been verified in this public feed. A planned check is not promised delivery.
Scheduled maintenance
Evidence state: stale scheduled pickup.
- Latest verified success included in this feed
- 25 Oct 2026, 14:30 UTC
- Last verified failure
- Not recorded. This does not mean no failures occurred.
- Success evidence readback
- 25 Oct 2026, 15:19 UTC
- Planned next evidence check
- Not recorded (not recorded)
A scheduled maintenance run completed and reconciled its project checkpoint. This is not a released checkpoint, successful source collection, complete research return, factual refresh, live heartbeat or sustained operation.
Checkpoint revision: 55fea19bfe85 (full revision in the public snapshot).
Recorded improvement work
These are recorded work states at the snapshot date, not live worker status. A completed implementation can still need outcome observation.
Better facility locations
Recorded state: active.
Check missing locations against usable originals.
Clearer development stages
Recorded state: ready.
Separate proposals, construction and supported operating evidence.
Public progress and evidence
Recorded state: waiting.
Verify the public replacement before retiring the private dashboard.
Evidence of visitor value
Recorded state: waiting.
Check consented journeys without treating clicks as successful access.
Inspect coverage, assessments and review history
Our aim is to support compute access, infrastructure and community decisions with inspectable evidence. Usefulness matters more than spending or construction volume.
Source records
234
Records can overlap. This is not a count of unique facilities.
Sourced coordinates
6 of 234
Other records may have an address or a country group.
Recorded lifecycle stage
66 of 234
A recorded field is not proof of present operation.
Local route checks
14 of 14
Software checks do not establish phone usability or visitor success.
The connected first edition
Explore the full lifecycle, find an access route, or read the briefing and three project evidence files. This selected cohort demonstrates the format; publication does not establish current project progress or global completeness.
Three promises
- Find a suitable route to compute, with conditions and an official next step.
- Understand what exists or is proposed in a place, with evidence and explicit unknowns.
- Understand what changed, with original dates and context.
What we can and cannot claim
The library includes 141 detailed site records, 84 provider records and 52 programme records. Regional coverage remains partial. See coverage and missing fields and the map and its country groups.
Latest recorded news date: 2026-11-12. Last recorded scoped news review: 2026-10-20. A review covers its stated sources and window, not the whole web. Site releases never refresh the evidence dates on individual records.
Visitor value: not yet measured. We have not established qualified audience, successful compute access or better community decisions. Requests, visits and clicks are different measures. No overall success score is assigned.
Desktop and phone acceptance: scoped observations only. These are checks of the named prior-release pages and widths, not whole-site usability, current-release acceptance or proof that a real person succeeded. Original observation times do not refresh with builds.
Desktop observations
- /dashboard: document fitted 1536px viewport widths at 16 Oct 2026, 22:43 UTC, revision
b6faa68f2847(historical, over 48 hours old). - /dashboard: document fitted 1536px viewport widths at 20 Oct 2026, 21:18 UTC, revision
b61da0a72cef(historical, over 48 hours old).
Phone observations
- /dashboard: document fitted 320, 390px viewport widths at 16 Oct 2026, 22:43 UTC, revision
b6faa68f2847(historical, over 48 hours old). - /finance: document fitted 390px viewport widths at 16 Oct 2026, 22:43 UTC, revision
b6faa68f2847(historical, over 48 hours old). - /services: document fitted 390px viewport widths at 16 Oct 2026, 22:43 UTC, revision
b6faa68f2847(historical, over 48 hours old). - /dashboard: document fitted 320, 390px viewport widths at 20 Oct 2026, 21:18 UTC, revision
b61da0a72cef(historical, over 48 hours old).
Coverage and the first reader edition
The 17 to 24 October 2026 first reader edition focuses on complete, understandable answers and a maintained update cycle. The reference and location observations below retain their original dates. Overall worldwide completeness: not established. The first independently checked external reference is deliberately bounded.
One publisher reference, not the whole market
Epoch AI major AI data-centre source records. Source updated 2026-10-12, retrieved 16 Oct 2026, 23:23 UTC, independently reviewed 16 Oct 2026, 23:25 UTC (reference observation is over 48 hours old).
- Existing exact identities represented
- 94 of 94 (100.0% of this reference only).
- Separately admitted canonical identities
- 1. Follow canonical record evidence; representation does not establish stage, capacity or availability.
- Reference records needing separate admission
- 0. An identified gap is not a populated record.
- Existing records with source address/location fields
- 80 of 93 (86.0%). Some values are localities or plus codes; these are not verified coordinates.
- Existing records whose source address is blank
- 13. Blank does not mean no facility or physical location.
- Fields included by the first batch
- 69 additional location fields. This is field enrichment, not 69 new facilities or a measured daily publication rate.
Exact source-record identities present in the unchanged estimate snapshot or independently admitted canonical identity records from this publisher reference. Not a worldwide census, deduplicated facilities, factual completeness or current estimates. Source-reported address/location fields, including localities and plus codes. Not sourced coordinates, verified physical identity, operation or availability.
Inspect the source records · Inspect attributed fields and dates · Inspect source-monitoring scope.
Are we keeping pace?
Material-change arrival rate, verified publication rate, publication delay and the required daily catch-up rate are not yet established. They require complete comparable source windows and actual review/release evidence. A build, imported field count or enabled schedule cannot establish market coverage or a deadline forecast. Other regional and topic reference sets still need independent assessment.
Inspect standing news intake, region/topic observations and uncovered areas. A scoped review is not whole-market coverage.
Are we becoming more useful?
Ten dimensions use the same 0-5 rubric. A score is a scoped, independently reviewed judgement, not a measurement of overall success. Unassessed means missing evidence, not zero. Privacy, source integrity and community agency remain required gates.
Visitor value
Discovery and usability: Unassessed
Can people complete the important journeys?
Evidence required: Observed phone and desktop task completion, failures and assistance. Agent checks remain separate from human success.
No calibrated numeric assessment is recorded. A dated unscored sample can provide evidence without a rating; inventory and passing tests alone cannot justify a score.
No unscored evidence sample is recorded for this dimension at this date.
Next useful action: Observe representative tasks on phone and desktop, recording failures and assistance.
Inspect the relevant evidence and limits. Follow-up uses the existing improvement programme.
Relevant reach: Unassessed
Are the people who need this finding it?
Evidence required: A defined relevant audience and compatible observations, with attribution and diagnostic exclusions stated.
No calibrated numeric assessment is recorded. A dated unscored sample can provide evidence without a rating; inventory and passing tests alone cannot justify a score.
No unscored evidence sample is recorded for this dimension at this date.
Next useful action: Establish a qualified audience baseline and attribution limits for the existing funded-training experiment.
Inspect the relevant evidence and limits. Follow-up uses the existing improvement programme.
Visitor value: Unassessed
Does this help someone find suitable compute or make an informed decision?
Evidence required: Voluntary task outcomes with successful respondents, all respondents, response coverage and selection limits. Clicks alone cannot answer this.
No calibrated numeric assessment is recorded. A dated unscored sample can provide evidence without a rating; inventory and passing tests alone cannot justify a score.
No unscored evidence sample is recorded for this dimension at this date.
Next useful action: Observe suitable next steps in a voluntary task sample, including unsuccessful responses and response coverage.
Inspect the relevant evidence and limits. Follow-up uses the existing improvement programme.
Information quality
Trust and accuracy: Unassessed
Can consequential claims be checked?
Evidence required: An independent sample of claims, including eligibility and availability, checked against original sources.
No calibrated numeric assessment is recorded. A dated unscored sample can provide evidence without a rating; inventory and passing tests alone cannot justify a score.
Unscored evidence sample: Six consequential source propositions
Three historical Finance accounts and one official-information link were checked. One earlier NVE source review remains historical; the ACCESS proposition was not rechecked in this sample.
- New scoped checks
- 4
- Named propositions
- 6
- Independently accepted scoped checks
- 4
- Historical Finance fact checks
- 3
- Official-information link checks
- 1
- Retained historical source reviews
- 1
- Not rechecked
- 1
Observed 31 Oct 2026, 22:59 UTC; independently reviewed 31 Oct 2026, 23:04 UTC; next scoped review 2026-11-07 (due; this dated sample remains historical). Source revision 280d54b7fb1641b10d6f15cdf2f94f5ebb85278d. These are evidence counts, with no numeric rating.
Next useful action: Audit consequential claims in a representative sample against their original sources.
Inspect the relevant evidence and limits. Follow-up uses the existing improvement programme.
Global coverage: Unassessed
Can we answer important questions across places and compute types?
Evidence required: A named question set across regions and audiences, including missed questions rather than record counts alone.
No calibrated numeric assessment is recorded. A dated unscored sample can provide evidence without a rating; inventory and passing tests alone cannot justify a score.
No unscored evidence sample is recorded for this dimension at this date.
Next useful action: Test important unanswered questions across regions, audiences and compute types.
Inspect the relevant evidence and limits. Follow-up uses the existing improvement programme.
Evidence depth: Unassessed
Is there enough evidence to take a useful next step?
Evidence required: Representative task records checked for conditions, constraints, uncertainty and official next steps.
No calibrated numeric assessment is recorded. A dated unscored sample can provide evidence without a rating; inventory and passing tests alone cannot justify a score.
Unscored evidence sample: Seven connected facility records
This records field presence at the named source revision. Gromstul retains its 25 October electrical-source check. Presence is not a fresh source review, current operation or available compute; prepared source candidates are excluded from these counts.
- Records checked for field presence
- 7
- Named detailed records
- 7
- Rows independently checked for presence
- 7
- Records with scoped grid evidence
- 1
- Records with scoped PUE evidence
- 0
- Records with scoped water evidence
- 0
- Records with scoped chip evidence
- 0
- Records with paired sourced coordinates
- 0
- Records with a recorded whole-site stage
- 4
Observed 31 Oct 2026, 23:10 UTC; independently reviewed 31 Oct 2026, 23:12 UTC; next scoped review 2026-11-07 (due; this dated sample remains historical). Source revision 280d54b7fb1641b10d6f15cdf2f94f5ebb85278d. These are evidence counts, with no numeric rating.
Next useful action: Identify the missing fact that prevents a useful next-step decision in sampled records.
Inspect the relevant evidence and limits. Follow-up uses the existing improvement programme.
Freshness: Unassessed
Is consequential information current within its stated scope?
Evidence required: Claim-level source dates, review scope, risk and expiry. A recent build is not a factual refresh.
No calibrated numeric assessment is recorded. A dated unscored sample can provide evidence without a rating; inventory and passing tests alone cannot justify a score.
Unscored evidence sample: Five overdue review obligations
Three historical Finance accounts, the KAUST information link and the full Services routing scope received separate reviews. Programme eligibility, deadlines and availability retain their own evidence dates. These counts do not measure continuous collection or automatic publication.
- Scoped reviews executed
- 5
- Previously overdue obligations
- 5
- Independently accepted scoped reviews
- 5
- Historical Finance fact reviews
- 3
- Link and routing reviews
- 2
- Unexecuted scopes in this sample
- 0
Observed 31 Oct 2026, 23:17 UTC; independently reviewed 31 Oct 2026, 23:22 UTC; next scoped review 2026-11-07 (due; this dated sample remains historical). Source revision 280d54b7fb1641b10d6f15cdf2f94f5ebb85278d. These are evidence counts, with no numeric rating.
Next useful action: Check volatile consequential claims against their original evidence and record a separate next review.
Inspect the relevant evidence and limits. Follow-up uses the existing improvement programme.
Ability to improve
Delivery reliability: Unassessed
Do useful changes reach the live site reliably?
Evidence required: Comparable release timings, failed handoffs, rework and owner interventions. Test counts are not delivered value.
No calibrated numeric assessment is recorded. A dated unscored sample can provide evidence without a rating; inventory and passing tests alone cannot justify a score.
No unscored evidence sample is recorded for this dimension at this date.
Next useful action: Evaluate the next three comparable releases using elapsed time, rework and owner effort.
Inspect the relevant evidence and limits. Follow-up uses the existing improvement programme.
Stewardship and work readiness: Unassessed
Does work remain owned, safe and aligned with the purpose?
Evidence required: Owned executable work, recovery evidence, privacy and source boundaries, community agency and owner effort.
No calibrated numeric assessment is recorded. A dated unscored sample can provide evidence without a rating; inventory and passing tests alone cannot justify a score.
No unscored evidence sample is recorded for this dimension at this date.
Next useful action: Check that consequential ready work has an owner, evidence, acceptance and a concrete return point.
Inspect the relevant evidence and limits. Follow-up uses the existing improvement programme.
Measurement and dashboard health: Unassessed
Can we trust the evidence used to choose work?
Evidence required: Reconciled evidence feeds, explicit denominators, reporting windows, exclusions and missingness.
No calibrated numeric assessment is recorded. A dated unscored sample can provide evidence without a rating; inventory and passing tests alone cannot justify a score.
No unscored evidence sample is recorded for this dimension at this date.
Next useful action: Reconcile each public checkpoint with its original evidence, window and missing observations.
Inspect the relevant evidence and limits. Follow-up uses the existing improvement programme.
Inspect the scoring rubric and definition
- Demonstrably absent or failing
- Basic capability with substantial gaps
- Partly useful with inconsistent coverage
- Reliable within an explicit scope
- Strong across representative needs
- Sustained independently checked performance against an explicit benchmark
Definition version 2. Scope, benchmark, checked/total sample, original observation, independent review and expiry are required. At expiry the current score is withdrawn; historical evidence is retained. A new build cannot renew it. Benchmark changes require a new version, never silent recalculation.
What we are learning
Evidence leads to a hypothesis, a bounded change, release verification and a later outcome check. A lesson is applied when a subsequent decision uses it. A released change is not proof of benefit.
Reliable journeys and faster useful delivery: Retain
- Hypothesis
- Removing observed journey failures and repeated rework should make useful improvements easier to deliver.
- Observation
- A reviewed decision is recorded. Its private supporting notes are excluded from this public summary.
- Evidence type
- Scoped browser check
- Changed approach
- The existing owner evaluates the recorded decision before the next increment.
- Next check
- 24 Oct 2026, 00:00 UTC (due for review). Check the next three comparable releases for elapsed time, rework and owner effort.
Observed 21 Oct 2026, 01:45 UTC; reviewed 21 Oct 2026, 01:47 UTC. No current priority promotion is inferred from this receipt.
Funded compute discovery: Inconclusive
- Hypothesis
- A clear funded-training journey should help researchers reach relevant programme evidence and an official next step.
- Observation
- A reviewed decision is recorded. Its private supporting notes are excluded from this public summary.
- Evidence type
- Insufficient outcome evidence
- Changed approach
- The existing owner evaluates the recorded decision before the next increment.
- Next check
- 24 Oct 2026, 00:00 UTC (due for review). Use the existing visitor review and a compatible settled audience window; keep benefit unknown until observed.
Observed 21 Oct 2026, 01:45 UTC; reviewed 21 Oct 2026, 01:47 UTC. No current priority promotion is inferred from this receipt.
Inconclusive results remain visible. Current independently reviewed change or reverse decisions may promote the affected ready work within its existing priority. Due daily reviews, dependencies, other owners and completed work stay protected. No second queue or schedule is created.
Whole-site review cycle
14 areas have a planned review cycle. Each check samples a named question or journey. A scoped review does not verify every fact in that area or establish worldwide completeness. Missing review evidence remains unknown; a build cannot refresh a review date.
Inspect review dates and planned cycles
| Area | Planned cycle | Last scoped review | Review state |
|---|---|---|---|
| News | Every 1 day | 2026-10-20 | Due for review. Gap found |
| Rules and controls | Every 3 days | 2026-10-20 | Due for review. Held: unresolved dependency |
| Verification and corrections | Every 3 days | 2026-10-20 | Due for review. Gap found |
| Delivery and recovery | Every 3 days | 2026-10-20 | Due for review. Gap found |
| Reach and measurement | Every 3 days | 2026-10-17 | Due for review. Gap found |
| Providers and access | Every 7 days | 2026-10-17 | Due for review. No change in reviewed scope |
| Places and policy | Every 7 days | 2026-10-17 | Due for review. No change in reviewed scope |
| Sites and communities | Every 7 days | 2026-10-17 | Due for review. Gap found |
| Supply chain | Every 7 days | 2026-10-17 | Due for review. No change in reviewed scope |
| Finance | Every 7 days | 2026-10-17 | Due for review. Gap found |
| Search and discovery | Every 7 days | 2026-10-17 | Due for review. Scoped correction recorded |
| Design and accessibility | Every 7 days | Not recorded | Not recorded |
| Research | Every 14 days | 2026-10-17 | Due for review. No change in reviewed scope |
| Guides and tools | Every 14 days | Not recorded | Not recorded |
Traffic development: not established. Search and audience evidence can guide improvements, but these review dates do not establish audience growth, returning visitors or successful compute access. Raw visitor analytics are excluded from this summary.
Evidence, definitions and corrections
Released site changes · Editorial method · Sources and licences · Published corrections · Download this public snapshot.
Snapshot source revision: ce3cd08585aa49ecbc4b5028f94b8da06c0c4018. Counts describe that source model; field presence does not establish accuracy. The public feed excludes private submissions, contacts, raw analytics and internal instructions.
Found a gap or a misleading claim? Suggest a sourced correction.