Nodes growing faster than utilisation is overcapacity.

Almost every DePIN network publishes supply — node counts, total capacity, provider tallies. Very few publish demand. We measure the ratio, on the networks where both halves are independently observable, and we hand you the command to check every number yourself.

All figures on this page were measured on our own infrastructure and re-verified against the upstream API at 2026-08-29 19:0xZ. Nothing here is modelled or estimated.

30.0% → 57.3%

Akash GPU utilisation, 2026-08-09 to 2026-08-28. Twenty days, twenty-seven points. Active GPUs went 127 → 248 (+95%) while total capacity went 423 → 433 (+2.4%). Demand nearly doubled; supply barely moved.

Verify that claim before you read another line

# Akash's own console API. active / total = current GPU utilisation.
curl -s https://console-api.akash.network/v1/network-capacity | jq '.resources.gpu'
# -> {"active":243,"pending":3,"available":177,"total":423}   => 57.4%

That is the entire product in one command. We did not invent the number, we noticed it, tracked it daily, and can prove when it moved.

What we currently measure

Network	Ratio	First	Latest	Change	Days
Akash	GPU	30.0%	57.3%	+27.3	20
Akash	CPU	20.1%	32.5%	+12.3	20
Akash	Memory	14.6%	24.8%	+10.2	18
Akash	Storage	8.3%	7.5%	−0.8	12
ThreeFold	CRU	14.2%	15.6%	+1.4	17
ThreeFold	MRU	13.7%	14.1%	+0.4	17
ThreeFold	SRU	8.1%	9.1%	+1.0	17
Storj	Node	27.2%	26.7%	−0.5	17

Eight live pairs across three networks, plus one held back pending a denominator check. This is a deliberately short list: a utilisation figure requires a numerator and a denominator from the same source at the same moment, and most networks publish only one of them. We would rather show you three we can prove than three hundred we cannot.

The engagement

Scope

You name the networks. We establish whether utilisation is observable at all — that answer alone has killed theses before a cheque was written.
Where it is observable, we stand up daily collection and report the ratio, the trend, and the denominator's behaviour. A denominator that moves is usually the real story.
Every figure ships with the upstream URL and the exact query. You can re-derive our work without us.
Where it is not observable, we say so in writing and explain what would have to change. We do not fill the gap with a model.

What you get

A written findings memo per network, with provenance for every number.
Daily machine-readable feed for the engagement period, via API or file.
The collectors themselves, documented, so the measurement outlives the engagement.

What this is not

Not price prediction. Not a token rating. Not a model output dressed as a measurement.
Not a subscription — it is scoped work with a named deliverable.

Honest limits, stated up front. Our continuous utilisation history begins 2026-08-09 and covers the networks in the table above. Longer histories exist for supply-side telemetry across 458 symbols, but utilisation specifically is young and narrow. If you need multi-year utilisation series today, we are not your vendor yet — and we will tell you that on the first call rather than the last.

Talk to us

Engagements are scoped per network set. If you are doing diligence on a DePIN position and want to know whether the demand side is real, start here.

Start a diligence engagement Or buy the data directly

Prefer to evaluate first? Every product is self-serve at /buy — $5 free credits, no card, no call.