Node health — what the headline counts hide

Findings from first-party endpoints · generated 2026-10-05T03:55:01Z

Four findings were tested against the underlying tables before anything was written here. Two survived, one survived only as a level, and one was refused. The refusals are at the bottom and they are the point: any scraper can produce numbers, and a competitor pointed at these same endpoints would have published all four.

1. Akash: the GPU market is tightening

Available GPUs fell -79.9% while active rose +164.6%. Idle supply is being absorbed.

Date	Total GPUs	Available	Active	Available share
2026-08-09	423	294	127	69.5%
2026-08-10	443	304	138	68.6%
2026-08-11	429	249	179	58.0%
2026-08-12	430	244	185	56.7%
2026-08-13	425	227	197	53.4%
2026-08-14	419	222	196	53.0%
2026-08-15	417	167	249	40.0%
2026-08-16	401	154	246	38.4%
2026-08-17	412	163	248	39.6%
2026-08-18	420	150	268	35.7%
2026-08-19	422	160	261	37.9%
2026-08-20	424	166	256	39.2%
2026-08-21	422	164	257	38.9%
2026-08-22	422	221	200	52.4%
2026-08-23	427	223	201	52.2%
2026-08-24	430	210	211	48.8%
2026-08-25	431	203	226	47.1%
2026-08-26	432	210	221	48.6%
2026-08-27	435	166	268	38.2%
2026-08-28	433	184	248	42.5%
2026-08-29	425	181	243	42.6%
2026-08-30	425	171	250	40.2%
2026-08-31	433	178	250	41.1%
2026-09-01	447	193	251	43.2%
2026-09-02	447	184	262	41.2%
2026-09-03	450	167	282	37.1%
2026-09-04	458	180	277	39.3%
2026-09-05	455	162	292	35.6%
2026-09-06	457	168	288	36.8%
2026-09-07	448	165	279	36.8%
2026-09-08	455	183	270	40.2%
2026-09-09	458	157	296	34.3%
2026-09-10	447	139	302	31.1%
2026-09-11	450	134	309	29.8%
2026-09-12	442	137	299	31.0%
2026-09-13	449	135	308	30.1%
2026-09-14	449	148	283	33.0%
2026-09-15	446	140	276	31.4%
2026-09-16	433	158	269	36.5%
2026-09-17	434	157	272	36.2%
2026-09-18	460	171	284	37.2%
2026-09-19	465	176	284	37.8%
2026-09-20	445	145	287	32.6%
2026-09-21	457	157	294	34.4%
2026-09-22	456	145	305	31.8%
2026-09-23	463	163	295	35.2%
2026-09-24	423	96	322	22.7%
2026-09-25	423	93	324	22.0%
2026-09-26	423	92	326	21.7%
2026-09-27	402	95	303	23.6%
2026-09-28	397	109	284	27.5%
2026-09-29	424	101	319	23.8%
2026-09-30	421	97	320	23.0%
2026-10-01	422	100	318	23.7%
2026-10-02	432	90	334	20.8%
2026-10-03	400	57	336	14.2%
2026-10-04	400	59	336	14.8%

Caveat, stated plainly: 57 daily observations is a direction, not a conclusion. A single large deployment would produce the same shape. We publish it because the direction is unambiguous and every point is checkable at console-api.akash.network — not because 4 points establish a trend.

2. Storj: the node count is inflated 4.0x

Bucket	Nodes	Share of total
Total (headline figure)	124,218	100%
Disqualified	82,376	66.3%
Active	31,161	25.1%

Quoting 124,218 nodes overstates the working network by 4.0x. Source: stats.storjshare.io.

Two things we will not claim. (a) The buckets are not a partition — active, disqualified, exited, offline and suspended sum to 126,203 against a stated total of 124,218, so they overlap by 1,985. We show only the two ratios that are well-defined against the total. (b) A tempting reading — “the network is accumulating dead nodes faster than live ones” — is not supported by our own data. Over 2026-08-06 to 2026-10-04, disqualified grew +2,141 while active grew -1,776. Active grew faster. We checked because the story was attractive.

3. ThreeFold: 82% of registered CPU is dark

396 nodes reporting up. 1,471 of 8,224 registered cores in use (17.9%). Source: gridproxy.grid.tf. See the utilisation report for the full capacity picture.

Refused — and why

“DefiLlama understates GEODnet revenue by 91%.” REFUSED.
GEODnet's API reports total_usage_revenue = 10,645,511; DefiLlama reports a 30-day revenue of 908,713. The ratio is 11.71x and is stable across all 7 days — which looks compelling and is meaningless as stated: GEODnet renders this field on its own homepage as “Current Annual Recurring Revenue” — an annualised run-rate — while DefiLlama's is a 30-day window. Annualise first and the two agree: 11.71 × 30/365 = 0.96, i.e. near parity. Dividing them as published manufactures a 91% “understatement” that does not exist.

This is the same error class our ontology already refuses for Akash (ACTIVE_LEASES / TOTAL_LEASES: a lifetime denominator yielding a meaningless 0.1%). Publishing it would have been a false public accusation against a third party, derived from a unit error.

Corrected 2026-09-06. This note previously called the field cumulative lifetime revenue and explained its mid-window fall as the series being “token-denominated and revalued daily”. Both readings were wrong: a cumulative total cannot fall, and dividing the value by GEOD spot swings 36% across the archive (42,559,689–57,920,771), so it is not a token quantity either. An annualised run-rate moves in both directions, which is what the archive shows. The refusal above stands; only its stated reason has changed.

“Filecoin revenue per miner rose 26% — oversupply is clearing.” REFUSED.
Active miners did fall (537→529). But the revenue increase behind that claim is a single day of a 24-hour series. Over the same window Filecoin's 30-day revenue fell 76.0% (2026-08-06 to 2026-10-04) — including a 17.6% drop in one day between 2026-08-07 and 2026-08-08, which is an upstream revision rather than anything organic. A per-unit ratio built on a 24h numerator over a 4-day window, when the 30-day series moves the other way, is a story rather than a finding.

Cross-network revenue-per-unit league tables. REFUSED.
A GEODnet weather station is not an Akash datacentre. The spread between them is arithmetically real and economically meaningless. Within-network direction is publishable; the cross-network level is not.

Capacity utilisation · All networks · Pricing · No predictive or performance claim is made anywhere on this site.