An Emerging Markets view of global cloud. How enterprises in India, Southeast Asia, the Middle East, and Latin America actually run cloud, and what the rest of the world should learn from them. Grounded in primary telemetry from 25 production cloud accounts.
Written and researched by ZopDev Research, Emerging Markets Cohort.
Read it front to back like a book, or open to the chapter your role needs. Page numbers refer to the print edition.
Written from close to the data, not from the coast that usually writes these.
Most State of Cloud reports are written from Silicon Valley, for Silicon Valley. This one is not.
They assume a dollar-denominated cloud bill, US labor costs, US regulatory pressures, and US patterns of adoption. Their case studies are American. Their benchmarks are American. Their advice is implicitly addressed to an American CIO. That coverage is fine. It is also incomplete.
India is now the third-largest cloud market in the world and the fastest-growing. The Middle East is investing over $50B in sovereign infrastructure. Southeast Asia is leapfrogging traditional IT entirely. Latin America runs cloud under currency conditions that fundamentally change the economics. None of this is visible in the existing reports. Not because the topics are unimportant, but because the analysts writing them are not close to the data.
We are. This edition is grounded in primary telemetry from 25 production cloud accounts across 13 enterprises, predominantly India-headquartered and India-presence multinationals, with operations spanning Asia-Pacific, the Middle East, Europe, and the Americas. Before opening a single dashboard we wrote ten falsifiable hypotheses. All ten were confirmed. The conventional wisdom about cloud is broadly correct. What is missing is the will to act on it.
The 16.9% waste finding holds whether you operate from Mumbai, Manchester, or Mountain View. How that waste accumulates is what differs by geography.
If you take only one statistic from this edition, take this one.
Sixteen point nine cents of every cloud dollar is recoverable waste, sitting in plain sight.1 For an enterprise spending $10M a year on cloud, that is $1.69M on the table, recoverable without architectural changes, vendor renegotiations, or workload repatriation.
The figure is not a survey estimate. It is a ratio: $375,217 of proven, annualized recoverable savings against $2,222,930 of annualized cohort spend. We counted only recommendations we could prove with high confidence, and excluded architectural rebuilds, repatriation, commitment renegotiations, and storage-tier changes. That is why we call it the floor, not the ceiling. The true recoverable figure is likely 30 to 40% of cloud spend once the excluded categories are in scope.
1Derivation, cohort, annualized: $375,217 recoverable over $2,222,930 spend equals 16.9%. Measured across a 69-day observation window and then annualized.
Derivation, cohort, annualized: $375,217 recoverable over $2,222,930 spend equals 16.9%. Measured across a 69-day observation window and then annualized.
The scale of the underlying observation is what makes the ratio defensible. We measured $420,554 of real spend across the window, $2.22M once annualized, spread over 25,225 live resources in 36 service types and 32 active cloud regions. The cohort spans FMCG, e-commerce, consumer internet, B2B logistics, sporting goods, global beverages, and creator-economy SaaS. This is telemetry from production accounts, not intent reported in a form. It cannot be biased by what a respondent wishes were true.
In Indian-rupee terms, the addressable waste at a single mid-sized enterprise can exceed ₹50 crore a year. The currency conversion sharpens the point rather than softening it, because the cloud bill is denominated in dollars while the revenue defending it is not. We return to that asymmetry in Chapter 07.
Roughly one dollar in every six of cloud spend is recoverable through hygiene alone, before any architectural change, renegotiation, or repatriation.
The industry solved seeing the waste. It never solved fixing it.
Detection has been solved. Execution has not. Cloud teams identify waste roughly 30 times faster than they act on it.2 The industry spent a decade building tools that surface recommendations, and almost none that carry them through. Fewer than 1% of detected waste signals in our cohort were ever formally applied through a remediation workflow. The work happens, or does not happen, in the cloud console, by hand, when someone finds the time.
Most of the waste that does disappear does not disappear on purpose. It ages out. Owners leave, projects end, teams reorganize, and the orphaned resource is eventually swept up in an unrelated cleanup. That is churn, not remediation. It means the improvement is not repeatable and cannot be planned around.
2Predicted ratio was 10 times. Observed was closer to 30 times, more extreme than the hypothesis. Confirmed across the full cohort.
Predicted ratio was 10 times. Observed was closer to 30 times, more extreme than the hypothesis. Confirmed across the full cohort.
The gap is not technical. It is human. Every enterprise in the cohort had the tooling. The Optimizer that cut daily spend in half in nine weeks used the same detection stack as the Drifter that grew ten percent a quarter without noticing. The difference was that someone owned the list, every day, and worked through it one recommendation at a time.
Three human forces keep the gap open. The first is ownership ambiguity: the account belongs to platform, the workload to the app team, the cost to finance, so the recommendation belongs to no one. The second is risk asymmetry, the fear of prod: deleting a snapshot carries a visible, personal risk of breaking something, while leaving it running costs a diffuse, unbilled-to-anyone amount, so the safe choice is always to leave it. The third is the missing incentive: no one is measured on resources deleted, so the work that saves money is the work that never gets prioritized.
The market for FinOps that detects is mature. The market that acts is wide open.
The same 16.9% of waste, arrived at by an entirely different road.
Currency exposure is a cloud risk, not a line item. Hyperscaler bills are denominated in US dollars. Revenue at most emerging-market enterprises is denominated in local currency. A 10% adverse currency movement, which has happened to the rupee, the real, and the rand more than once in the last five years, is a 10% cloud cost increase.3 No new resource is provisioned. The bill simply grows, and the finance team that signed off on the budget in January is defending a different number by June.
That single fact rewrites the playbook. Multi-year reserved-instance commitments look attractive on a US dashboard because they lock in dollar pricing, and risky on an Indian dashboard because they lock in dollar exposure. Repatriation conversations begin earlier in emerging markets, because the local-currency math turns favorable sooner. And sovereign cloud offerings denominated in local currency command a premium even at a higher list price, because the currency hedge is itself worth paying for. The more aggressive sovereign-cloud activity is happening outside Europe, backed by the Middle East's investment of over $50B in local infrastructure.
3A treasury risk, not a tech line item. The same commitment that de-risks a US bill adds risk to an emerging-market one.
A treasury risk, not a tech line item. The same commitment that de-risks a US bill adds risk to an emerging-market one.
Labor economics bend the AI story too. Compute pricing is roughly global. Engineering pricing is not. The fully-loaded cost of one US developer is close to the cost of seven engineers in India. When a task can be solved with more people or more GPUs, the tradeoff a US team makes and the tradeoff an Indian team makes are not the same tradeoff, even facing identical cloud prices.
A 10% currency swing is a 10% cost increase, without a single new resource being provisioned.
Commentary from readers who saw their own bill in the findings. Attributions are anonymized by role and organization profile.
Reader commentary · not findings of the report
“We had every recommendation this report describes, sitting in a dashboard nobody owned. The 30-times number is the most honest thing I have read about our own operation.”
“The currency chapter changed how our board talks about reserved instances. A dollar commitment is a treasury decision, not a discount.”
“Finally a report that counts what it can prove and calls the rest the ceiling.”
“The fear-of-prod section is uncomfortably accurate. Our engineers are not lazy. They are correctly avoiding a risk no one rewards them for taking.”
“One engineer costs what seven cost here. That one line reframed our entire build-versus-buy conversation on AI.”
16.9% of every cloud dollar is recoverable, proven and before any architectural change, with the true figure likely 30 to 40% once excluded categories are counted. Fewer than 1% of detected waste is formally remediated, because detection is solved and action is not. A 10% currency swing is a 10% cost increase for enterprises whose bill is in dollars and whose revenue is not. And 10 of 10 hypotheses, all written before a single dashboard opened, were confirmed against telemetry from 13 firms across four-plus regions. The conventional wisdom is right. What is missing is the will to act on it.
The State of Cloud 2026 is published by ZopDev Press as a collectible annual. Read it on the web, take the PDF into the boardroom, or request the print-ready edition for your team.
The full interactive edition, free to read and share. No signup, no paywall. The version you are reading now.
Open the reader →A print-formatted PDF with the title page, contents, and margin notes intact. Built for circulating to a leadership team.
Download the PDF →A press-quality file for your own run, on request. The complete edition with appendices and the full methodology.
Request access →ZopDev Press is the publishing imprint of ZopDev, the multi-cloud company behind ZopNight, ZopDay, and ZopCloud. We publish research meant to be argued with, not archived.
The most useful thing you can do with this edition is contest it. We published the full methodology so you can replicate, challenge, and improve on the findings.