ZopDev · State of Cloud 202601 / 12
↓ PDF

ZopDev Research · Drawing Set 2026.06

The State
of Cloud
2026

An architecture dossier drawn from live telemetry: 25 production cloud accounts · 13 enterprises · India, SEA, Middle East, LatAm. Text is annotation. The drawing carries the story.

$2.22Mannualized spend analyzed
25,225live resources surveyed
16.9%provably recoverable waste
AWS 65.5% OF SPEND ap-south-1 us-east-1 eu-west-1 AZURE 25.7% OF SPEND GCP 8.7% 6 OF 13 SINGLE-CLOUD AWS FIG. 0 · COHORT ESTATE, ELEVATION · NOT TO SCALE GENUINE MULTI-CLOUD: 1 OF 13 N
An Emerging
Markets Viewfield-verified · june 2026
Open the drawing set Download PDF
Sheet01 of 12ScaleNTSRev2026.06Drawn byZopDev Research
02

orientation · rev 2026.06

Sheet index & survey notes.

This excerpt plots the three most diagrammable chapters of the full report — the footprint, the waste, and the architecture — as numbered plates. Read the drawings first; the prose is annotation.

NoChapterStatus
ESExecutive Summary — cloud stopped being new; the bill is still a mysteryFull edition
01Hypothesis Scoreboard — 10/10 pre-registered hypotheses confirmedFull edition
02The Defensible Number — 16.9% provable; true figure likely 30–40%Full edition
03The State of Enterprise Cloud — spend, footprint, trajectoryPlates 01–02
04Where the Waste Lives — 87% forgotten, not strategicPlate 03
05The Detection-to-Action Gap — 30× faster to find than to fixFull edition
06The Architectural Picture — sprawl is the new technical debtPlates 04–05
07An Emerging Markets View — currency, sovereignty, labor economicsFull edition
08Predictions for 2027 — falsifiable, dated, gradeableFull edition
09The GPU Shift — the same anti-patterns, three more zerosFull edition
CLClosing — three trends, three counter-trends, one betFull edition
25production accounts
13enterprises · 4+ regions
69days of telemetry
32cloud regions covered
10/10hypotheses confirmed

Survey instruments: native AWS / Azure / GCP resource-discovery and cost APIs, plus a 315+ rule recommendation engine, applied uniformly. Window: March 20 – May 27, 2026. Published June 2026.

03

the state of enterprise cloud

Three views: spend, footprint, trajectory.

Spend by provider tells a different story than account count. AWS still takes two-thirds of every cloud dollar — and account count and spend are not the same shape.

PLATE 01· Cloud Footprint Map
AWS 53% OF ACCOUNTS · 6 OF 13 SINGLE-CLOUD 65.5% EC2 · RDS EKS S3 EBS · SNAPSHOTS AZURE 28% OF ACCOUNTS 25.7% GCP 19% OF ACCOUNTS 8.7% 1 OF 13 65.5¢ OF EVERY CLOUD DOLLAR · COHORT TELEMETRY 1 2 3

Areas proportional to annualized spend share · cohort telemetry, 25 accounts

Plate01Scale1¢ : 6.6px²Rev2026.06Drawn byZopDev Research

Footprint itself is a choice — most enterprises have not made it consciously. Two of the four patterns below reflect deliberate architecture. Two reflect accident.

Table 3.3 · Footprint patterns · anonymized
Footprint patternExampleRegions$ / region / mo
Global sprawl, one tenantGlobal sporting goods · Azure32~$60
Concentrated productionGlobal CPG · Azure3~$5,900
Single-region monolithConsumer internet · AWS2~$12,500
Spread without intentE-commerce · AWS19~$560
3.1

the bimodal economy

Three archetypes, three trajectories.

The "average enterprise cloud bill" is a meaningless number. In the same quarter, one customer halved spend while another quadrupled it. The story is in the spread.

PLATE 02· Archetype Trajectories
SPEND · INDEXED RUN-RATE OBSERVATION WINDOW · 69 DAYS ×4.1 THE SCALER · ~42% · $1,429 → $5,918/WK · 6 WKS +10% THE DRIFTER · ~50% · $5,787 → $6,361/WK · 8 WKS −50% THE OPTIMIZER · ~8% · $1,565 → $786/DAY · 69 DAYS THE SPREAD 1 2 3

Real customers, observed run-rate · curves indexed, not to a common $ scale

Plate02ScaleNTSRev2026.06Drawn byZopDev Research

Which trajectory are we on — and is it the one we chose?

04

where the waste lives

87% of cloud waste isn't strategic. It's forgotten.

Vendor narratives focus on Reserved Instances, Savings Plans, Spot — only 6.3% of what we flagged. The real waste is things nobody remembers creating.

PLATE 03· Waste Topology
VPC · PRODUCTION ACCOUNT · THE PART EVERYONE REVIEWS EC2 FLEET 7 instances · 6 running RDS db.t3.small · prod EKS CLUSTER 4 node groups S3 standard tier LIVE WORKLOADS · OWNED · MONITORED ORPHAN SNAPSHOTS 3,807 in one account · charged forever UNATTACHED EBS ×8 each leaks ~$2,500 / yr IDLE NAT GATEWAY one per Databricks workspace ABANDONED IP · AMI stopped EC2 still billing EBS 87.4% OF ALL FLAGGED ITEMS · ~5 MIN FIX EACH DISCOUNT 6.3% · RIGHTSIZING 3.7% · IDLE 1.3% · SCHEDULE 0.9% 1 2 3 4

Share of flagged items, cohort-wide · 315+ detection rules applied uniformly

Plate03ScaleNTSRev2026.06Drawn byZopDev Research
4.1

the hall of shame

Fix ten things. Recover 80% of the waste.

None of these are strategic decisions to revisit. They are hygiene to install. If your platform team cannot recite this list from memory, you have a 16.9% problem.

Table 4.2 · Anti-pattern schedule · universality & annual stake
#Anti-patternUniversality$ / yr at stake
01Orphan EBS snapshots — no source volume, charged forever100% of AWS users$80K+
02Unattached EBS volumes — single volumes leak $2,500/yr100% of AWS users$25K+
03Premium SSD in non-prod Databricks (Azure)100% of Databricks on Azure$66K+
04EC2 not covered by Savings Plans~90% of prod fleets$50K+
05Single-AZ production RDS — not cost, riskEvery AWS enterpriseResilience risk
06Dev/test workloads with no schedule100% of dev environments$40K+
07Windows VMs not using Azure Hybrid BenefitCommon in regulated industries40% on license
08On-demand for stateless / batch workloadsCommon in cloud-native shops$30K+
09x86 where Graviton / ARM would work80% of compute fleets$25K+
10Stopped instances still incurring EBS chargesFound in every account$5K+
10 fixes → 80%of recoverable waste · before any architecture change
06

the architectural picture

Resource sprawl is the new technical debt.

For CTOs and cloud architects. The visible compute layer of any production account is dwarfed by its metadata — and the metadata, unlike the compute, has no lifecycle.

PLATE 04· The Metadata Iceberg — Section View
WATERLINE · WHAT THE BILL-REVIEW MEETING SEES COMPUTE EC2 · the visible layer EBS SNAPSHOTS 14× the live volumes · 2.3 snapshots / volume / yr CLOUDWATCH ALARMS 22 per EC2 instance · 6,922 in one account, mostly unmonitored IAM ROLES · POLICIES · LOG GROUPS 3 IAM roles per workload · no deprecation window, no owner METADATA GROWS · WORKLOAD FLAT THE BILL GROWS LINEARLY WITH METADATA, EVEN WHEN WORKLOAD IS FLAT. 1 2 3

Section through a typical production AWS account · strata widths ∝ item count

Plate04ScaleNTSRev2026.06Drawn byZopDev Research
6.1

the fragmentation problem (azure)

Consolidators vs Fragmenters.

One customer ran 33 distinct Azure subscriptions in a single tenant. Another ran 3 for comparable spend. The first is governance debt waiting to compound. The second is governance done right.

Consolidators

3–5 large subscriptions

Cost / subscription / mo
~$15–20K
Governance
Tractable
Cleanup
Feasible

Fragmenters

20–35 small subscriptions

Cost / subscription / mo
~$500–1,500
Governance
Scales with team count
Cleanup
Hopeless

The most visible symptom of fragmentation: Databricks workspace proliferation — each workspace spinning up its own NAT gateway and Premium SSD scratch volumes. Confirmed across every Databricks-on-Azure enterprise in the cohort.

6.2

resilience theater

Every AWS enterprise had at least one Single-AZ production database.

The reason is never "we made a deliberate choice." It is always "we ran the migration script in 2021 and never went back."

PLATE 05· Single-AZ Risk — Plan View
REGION · ap-south-1 AVAILABILITY ZONE A APP TIER PROD RDS MultiAZ = false AVAILABILITY ZONE B NO STANDBY REPLICA · EMPTY REPLICATION · NOT CONFIGURED AZ FAILS → MULTI-HOUR OUTAGE WITH MULTI-AZ → ~5-MIN FAILOVER 1 2 3

Hatched region indicates unmitigated availability hazard · rule RC-005

Plate05ScaleNTSRev2026.06Drawn byZopDev Research
R

survey summary · as-measured

The dossier, in four dimensions.

16.9%provably recoverable waste

Roughly $1 in every $6 of cloud spend, recoverable through hygiene alone — before any architectural change. The true figure is likely 30–40%; we refuse to claim what we cannot prove.

87%of waste is non-strategic orphans

Idle compute, unattached volumes, forgotten snapshots, NAT gateways. Fixing the top ten anti-patterns recovers ~80% of the total.

65.5%of cohort spend sits on AWS

Genuine multi-cloud is 1 in 13. Footprint is mostly accidental — and every AWS enterprise carried at least one Single-AZ production database.

10 / 10pre-registered hypotheses confirmed

Written before a single dashboard was opened. The conventional wisdom about cloud is broadly correct — what's missing is the will to act on it.

The boardroom question is not "can we cut cloud spend?" It is "why haven't we already?"

Closing · issued for review

The most important thing you can do with this drawing set is challenge it.

We published the full methodology so you can replicate, contest, and improve it. 13 enterprises is a small sample; when we say 16.9% we mean the central estimate — the true industry-wide value is probably in the 12–22% range. Challenges: contact@zop.dev, 14-day response window.

One bet, willing to be publicly wrong: by the 2027 edition, more than half of this cohort crosses into Optimizer territory — not because tooling improved, but because the next downturn makes cloud cost a board priority.

Talk to ZopDev Download PDF

The State of Cloud 2026 · An Emerging Markets View · Edition 1 · June 2026

Sheet12 of 12 Rev2026.06 StatusIssued for review Drawn byZopDev · Industry Research