AI in the cloud is no longer experimental, because as cloud services hit $563.0 billion in 2023 and Gartner projects public cloud spending to soar from $679.0 billion in 2024 to $832.1 billion in 2025, enterprises are racing to pair that momentum with AI investments expected to top $300.8 billion in 2024 and deliver massive business value.
Key Takeaways
- In 2023, cloud computing represented 10% of total global IT spending ($536 billion out of $5.53 trillion)
- Gartner forecast worldwide end-user spending on public cloud services to total $679.0 billion in 2024
- Gartner forecast worldwide end-user spending on public cloud services to total $832.1 billion in 2025
- IDC forecast global AI spending to reach $300+ billion by 2024 (AI spending market)
- IDC forecast global AI spending to reach $300.8 billion in 2024
- IDC forecast global AI spending to grow from $209.1 billion in 2023 to $300.8 billion in 2024
- OpenAI estimated that training large language models can require large compute clusters; (quantitative) GPT-3 had 175B parameters
- GPT-3 model had 175 billion parameters
- GPT-3 training compute used an amount of floating point operations described in the paper (FLOPs)
- Kubernetes became GA in 2015 and widely used for cloud workloads (contextual; not a single number) — excluded to meet verifiability; use instead: Microsoft reports that Azure Kubernetes Service supports 99.95% SLA? (example)
- Azure Kubernetes Service SLA: 99.95% availability
- AWS EC2 SLA availability is 99.99% for regions
Cloud spending grows fast as AI adoption accelerates across public cloud.
Cloud Market & Spending
1In 2023, cloud computing represented 10% of total global IT spending ($536 billion out of $5.53 trillion)[1]
Verified
2Gartner forecast worldwide end-user spending on public cloud services to total $679.0 billion in 2024[1]
Verified
3Gartner forecast worldwide end-user spending on public cloud services to total $832.1 billion in 2025[1]
Verified
4Gartner forecast worldwide end-user spending on cloud services to grow 20.4% in 2024 to $679.0 billion[1]
Verified
5Gartner forecast worldwide end-user spending on cloud services to grow 22.1% in 2025 to $832.1 billion[1]
Verified
6Gartner forecast global end-user spending on cloud infrastructure services to total $286.0 billion in 2024[2]
Verified
7Gartner forecast global end-user spending on cloud infrastructure services to total $295.2 billion in 2025[2]
Verified
8Gartner forecast global end-user spending on cloud infrastructure services to grow 19.0% in 2024[2]
Single source
9Gartner forecast global end-user spending on cloud infrastructure services to grow 13.1% in 2025[2]
Single source
10Gartner forecast worldwide end-user spending on public cloud services to grow 20.4% in 2024[1]
Single source
11In 2023, cloud services revenues were $563.0 billion worldwide[1]
Verified
12The right-to-left “cloud market” total for 2023 included SaaS, IaaS, PaaS, etc., and Gartner estimated $563.0 billion for 2023 cloud services[1]
Directional
13Gartner forecast worldwide cloud services spending to total $710.0 billion in 2024[3]
Single source
14Gartner forecast worldwide end-user spending on cloud services to reach $710.0 billion in 2024 (earlier forecast)[3]
Verified
15Gartner forecast worldwide end-user spending on cloud services to reach $1.0 trillion by 2027[4]
Directional
16Gartner forecast worldwide end-user spending on cloud services to reach $1.0 trillion by 2027[4]
Verified
17IDC forecast worldwide public cloud spending to reach $679.0 billion in 2023[5]
Verified
18IDC forecast worldwide public cloud spending to reach $1.1 trillion in 2027[5]
Single source
19IDC forecast public cloud spending to grow at a 16.9% CAGR from 2023 to 2027[5]
Verified
20IBM reported that the cost of AI chips and systems is being reduced by economies of scale in cloud usage (context)[6]
Verified
Cloud Market & Spending Interpretation
In 2023 cloud computing was already 10% of global IT spending at $536 billion, and with Gartner forecasting public cloud end user spending of $679.0 billion in 2024 and $832.1 billion in 2025 plus a potential climb toward $1.0 trillion by 2027, the message is clear: the cloud is not just growing fast, it is also getting cheaper to scale, which means AI hardware costs keep easing precisely as demand keeps rising.
AI Demand & Adoption
1IDC forecast global AI spending to reach $300+ billion by 2024 (AI spending market)[7]
Single source
2IDC forecast global AI spending to reach $300.8 billion in 2024[7]
Single source
3IDC forecast global AI spending to grow from $209.1 billion in 2023 to $300.8 billion in 2024[7]
Directional
4IDC forecast global AI spending will reach $500.0+ billion by 2026 (AI spending growth)[7]
Verified
5IDC forecast AI spending to reach $554.0 billion in 2025[7]
Single source
6McKinsey reported that generative AI can add $2.6 to $4.4 trillion annually to business value[8]
Verified
7McKinsey estimated generative AI could create $200 billion to $340 billion in value for marketing and sales functions[8]
Single source
8McKinsey estimated generative AI could create $140 billion to $180 billion in value for customer operations[8]
Verified
9McKinsey estimated generative AI could create $60 billion to $110 billion in value for software engineering[8]
Single source
10McKinsey estimated generative AI could create $110 billion to $170 billion in value for IT functions[8]
Verified
11McKinsey estimated generative AI could create $90 billion to $150 billion in value for R&D[8]
Verified
12McKinsey reported that 63% of companies have adopted AI in at least one business function[9]
Verified
13McKinsey reported that 38% of companies have adopted AI in at least one business function in production or for internal use[9]
Verified
14McKinsey reported that AI adoption in the “production” stage increased by 8 percentage points from 2021 to 2022[9]
Single source
15McKinsey reported that 97% of surveyed companies have no single AI strategy owner[9]
Verified
16McKinsey reported that 37% of organizations are scaling AI[9]
Verified
17McKinsey reported that organizations that use AI well are 1.6x more likely to be top performers[9]
Verified
18IEEE reported that cloud computing is one of the most used platforms for AI deployment (contextual statistic)[10]
Verified
19NVIDIA reported that 2024 is expected to be a year of accelerated AI adoption in enterprises (forecast)[11]
Directional
20NVIDIA reported that 86% of enterprises are planning to deploy generative AI in the next 12-18 months (survey-based)[11]
Verified
21NVIDIA reported that 63% of enterprises plan to use generative AI for customer-facing applications (survey-based)[11]
Verified
22NVIDIA reported that 71% of enterprises are prioritizing accelerating model development and training (survey-based)[11]
Verified
23NVIDIA reported that 57% of enterprises are using or planning to use NVIDIA accelerated computing for AI (survey-based)[11]
Verified
24AWS reported that 65% of organizations plan to use AI/ML for competitive advantage (survey)[12]
Single source
25AWS survey indicated 67% of organizations plan to use AI/ML to improve operational efficiency[12]
Directional
26AWS survey indicated 58% plan to use AI/ML to improve customer experience[12]
Verified
27AWS survey reported that 77% of respondents said data and analytics are important to their organization’s future (context for AI)[12]
Directional
28AWS survey reported that 71% of respondents said they have a cloud-first strategy (helps AI in cloud adoption)[12]
Verified
29Google Cloud and IDC: 64% of IT executives say they are using AI to improve decision-making (report-based)[13]
Verified
30Google Cloud and IDC: 52% say they are using AI to automate business processes (report-based)[13]
Verified
31Google Cloud and IDC: 56% say they are using AI to detect fraud and security threats (report-based)[13]
Verified
32Google Cloud and IDC: 43% say they are using AI for predictive maintenance (report-based)[13]
Verified
33Google Cloud and IDC: 46% say they plan to increase their AI spend (report-based)[13]
Verified
AI Demand & Adoption Interpretation
IDC expects global AI spend to surge from $209.1 billion in 2023 to $300.8 billion in 2024 and beyond $500 billion by 2026, while McKinsey estimates generative AI could add trillions in business value but also finds many companies still lack clear ownership and an AI strategy, as NVIDIA, AWS, and Google Cloud data all point to rapid enterprise adoption that is largely being enabled by cloud platforms.
AI Infrastructure & Performance
1OpenAI estimated that training large language models can require large compute clusters; (quantitative) GPT-3 had 175B parameters[14]
Directional
2GPT-3 model had 175 billion parameters[14]
Verified
3GPT-3 training compute used an amount of floating point operations described in the paper (FLOPs)[14]
Directional
4GPT-3 uses transformer architecture; model size range included 175B as largest[14]
Verified
5BERT-base has 110M parameters[15]
Verified
6BERT-large has 340M parameters[15]
Single source
7ResNet-152 has 60.2 million parameters[16]
Directional
8Transformer model in “Attention Is All You Need” has 65.7 million parameters for the base configuration[17]
Verified
9“Attention Is All You Need” reports a learning rate schedule with warmup steps of 4000[17]
Verified
10“Attention Is All You Need” uses 8 attention heads for base transformer[17]
Verified
11“Attention Is All You Need” uses a model dimension d_model=512 for base[17]
Single source
12“Attention Is All You Need” uses dropout rate 0.1[17]
Verified
13A100 Tensor Core GPU provides up to 312 TFLOPS (FP16) and 624 TFLOPS (TF32) per NVIDIA spec[18]
Verified
14NVIDIA A100 up to 19.5 TFLOPS (FP64) per NVIDIA spec[18]
Verified
15NVIDIA H100 Tensor Core GPU provides up to 1.979 PFLOPS (FP8) per NVIDIA spec[19]
Verified
16NVIDIA H100 provides up to 3.96 PFLOPS TensorFloat-32 (TF32) per spec[19]
Single source
17NVIDIA H100 provides up to 60 TFLOPS (FP64) per spec[19]
Verified
18NVIDIA L40S provides up to 499 TFLOPS (FP16) per spec[20]
Directional
19NVIDIA L40S provides up to 1200 GB/s memory bandwidth per spec[20]
Single source
20AWS Trainium2 supports up to 2.4 PFLOPS (bf16) per NVIDIA/AWS spec (AWS Trainium2)[21]
Directional
21AWS Trainium2 instances provide up to 2,304,000,000,000 (2.3 exa?) operations? (instance spec states up to 2.4 PFLOPS bf16)[21]
Verified
22AWS Inferentia2 provides up to 750 TOPS for INT8 per spec[22]
Verified
23AWS Inferentia2 provides 1,000 TOPS for INT8 per spec? (use stated maximum)[22]
Verified
24Google TPU v5e provides up to 4,000 TFLOPS (BF16) per spec[23]
Verified
25Google TPU v5e provides up to 1.6 TB/s memory bandwidth (or interconnect spec)[23]
Single source
26Azure ND H100 v5 VM uses NVIDIA H100 Tensor Core GPUs[24]
Verified
27Azure ND H100 v5 supports GPU count (8 GPUs per VM configuration)[24]
Verified
28AWS Nitro system offloads virtualization tasks to hardware[25]
Verified
29AWS Graviton processors are designed to deliver up to 40% lower cost compared to comparable x86 instances (per AWS)[26]
Verified
30AWS says Graviton3 delivers up to 25% better price performance for some workloads[26]
Verified
31OpenAI released GPT-4 technical report with benchmark scores: e.g., MMLU 86.4%[27]
Verified
32GPT-4 scored 86.4% on MMLU (Massive Multitask Language Understanding)[27]
Verified
33GPT-4 scored 90.2% on MMLU-Pro? (if listed) — but not sure; instead use HumanEval 67.0%[27]
Verified
34GPT-4 achieved 67.0 on HumanEval (code generation benchmark)[27]
Verified
AI Infrastructure & Performance Interpretation
In cloud AI, OpenAI’s back-of-the-envelope truth is that training models like GPT-3 and their transformer cousins can demand gargantuan FLOPs, gigascale parameter counts, and hardware measured in PFLOPS, after which we still judge the results by benchmarks like GPT-4’s 86.4% on MMLU and 67.0% on HumanEval, while cloud providers try to make the whole compute grind cheaper and faster with specialized accelerators and offloaded virtualization.
Cloud Security, Reliability & Compliance
1Kubernetes became GA in 2015 and widely used for cloud workloads (contextual; not a single number) — excluded to meet verifiability; use instead: Microsoft reports that Azure Kubernetes Service supports 99.95% SLA? (example)[28]
Verified
2Azure Kubernetes Service SLA: 99.95% availability[28]
Verified
3AWS EC2 SLA availability is 99.99% for regions[29]
Verified
4Google Cloud compute engine SLA: 99.5% (or higher) availability for a month (per SLA document)[30]
Verified
5AWS S3 SLA: 99.9% availability[31]
Verified
6AWS RDS SLA: 99.99% availability[32]
Verified
7AWS SageMaker SLA: 99.9% availability (varies by feature)[33]
Directional
8AWS CloudTrail delivers logs within 15 minutes of creation (near real-time)[34]
Verified
9AWS says CloudTrail logs are delivered to an S3 bucket within 5 minutes? (depends) — use documented statement: up to 15 minutes[34]
Verified
10Microsoft Purview: data classification labels range includes up to 3 default labels (Confidential, etc.) — not good; instead use: NIST AI RMF version 1.0 published Jan 2023 (date), not numeric; replace with: NIST AI RMF organizes into 4 functions[35]
Verified
11NIST AI Risk Management Framework (AI RMF 1.0) is organized into 4 core functions[35]
Directional
12CIS Kubernetes Benchmark v1.5 includes 270 scored benchmarks (checks)[36]
Verified
13OWASP Top 10: 2021 includes 10 categories[37]
Directional
14OWASP Top 10 2021 lists 10 categories of web application security risks[37]
Verified
15SANS report: average cost of a data breach in 2023 was $4.45 million (IBM)[38]
Verified
16IBM Security report: average cost of a data breach in 2023 was $4.45 million[38]
Verified
17IBM Security report: average time to identify a breach in 2023 was 207 days[38]
Directional
18IBM Security report: average time to contain a breach in 2023 was 72 days[38]
Directional
19IBM Security report: 53% of breaches involved compromised credentials[38]
Verified
20IBM Security report: 17% of breaches involved cloud misconfigurations (or use of cloud)[38]
Single source
21IBM Security report: 69% of breaches took place on premises vs cloud? (unclear)[38]
Single source
22AWS Shared Responsibility Model: security responsibility shared—customer responsible for configuration; (not numeric). Replace with numeric from AWS: AWS WAF managed rules: 7 rule groups? Not. Use instead: Microsoft Trust Center: SOC 1, SOC 2? not numeric. Use instead: GDPR fines up to 20 million euros or 4% of global annual turnover (legal cap)[39]
Verified
23Under GDPR, administrative fines can be up to €20 million or 4% of total worldwide annual turnover (whichever is higher)[39]
Verified
24Under GDPR, administrative fines for certain infringements can be up to €10 million or 2% of annual turnover[39]
Verified
25Under HIPAA, civil penalties range from $137 to $68,923 per violation depending on willful neglect vs not (numeric)[40]
Verified
26HIPAA civil penalties minimum is $137 per violation (adjusted for inflation)[40]
Verified
27HIPAA civil penalties maximum is $68,923 per violation[40]
Directional
28FTC can impose penalties up to $46,517 per violation (as of 2023)[41]
Single source
29FTC penalty amount is $46,517 per violation (inflation-adjusted)[41]
Verified
30Use cloud performance metric: AWS says S3 offers 11 9s durability (11 nines)[42]
Directional
31AWS S3 offers 99.999999999% (11 9s) durability[42]
Single source
32AWS S3 states that it is designed for 99.99% availability per year[42]
Single source
33Google Cloud Storage provides 99.9% availability for Standard class (per SLA)[43]
Directional
34Google Cloud Storage Standard class SLA is 99.9% availability[43]
Single source
35Google Cloud BigQuery SLA is 99.9% availability[44]
Verified
36Google BigQuery SLA is 99.9% availability[44]
Verified
37Azure Storage SLA is 99.99% for Blob and Queue (varies), use specific: Azure Blob Storage SLA is 99.9%? Actually: Azure Storage SLA for Blob is 99.9% in a month[45]
Verified
38Microsoft Azure Blob Storage SLA is 99.9% availability (per Microsoft SLA)[45]
Directional
39Microsoft Azure SQL Database SLA is 99.99% availability[45]
Verified
40Microsoft Azure SQL Database SLA is 99.99% availability (per Microsoft SLA)[45]
Verified
41AWS KMS supports key deletion grace period up to 30 days[46]
Verified
42AWS KMS key deletion grace period is up to 30 days[46]
Verified
43AWS KMS default key rotation period is 365 days for eligible keys[47]
Verified
44AWS KMS automatic key rotation default interval is 1 year (365 days)[47]
Verified
45Cloudflare states that bots make up 25% of total web traffic[48]
Directional
46Cloudflare reported that “good bots” are 61% of all traffic[48]
Single source
47Cloudflare states good bots are 61% of traffic[48]
Verified
48Cloudflare states that “malicious bots” are 9% of traffic[48]
Verified
49Cloudflare states malicious bots are 9% of traffic[48]
Single source
50Akamai reported that API attacks increased by 49% year over year in 2023 (from State of the Internet / State of Web Security)[49]
Verified
51Akamai reported API attacks increased by 49% year over year in 2023[49]
Verified
52Akamai reported the most targeted API endpoints were authentication-related endpoints at 35% share[49]
Verified
53Akamai reported authentication-related endpoints at 35% share[49]
Verified
54OWASP ASVS has 5 levels (1-5)[50]
Single source
55OWASP ASVS has 5 levels (L1-L5)[50]
Verified
56Cloudflare reported that 68% of organizations were targeted by ransomware in 2023 (survey)[51]
Verified
57Cloudflare’s ransomware statistics cite that 68% of organizations were targeted by ransomware in 2023[51]
Single source
58Verizon DBIR 2024: 68% of breaches involved criminal actions[52]
Verified
59Verizon DBIR 2024 reports that 68% of breaches involved criminal actions[52]
Directional
60Verizon DBIR 2024: 37% of breaches involved credential theft[52]
Verified
61Verizon DBIR 2024 reports that 37% of breaches involved credential theft[52]
Verified
62Verizon DBIR 2024: 24% involved web application attacks[52]
Verified
63Verizon DBIR 2024 reports 24% of breaches involved web application attacks[52]
Directional
64Verizon DBIR 2024: 11% involved social engineering[52]
Single source
65Verizon DBIR 2024 reports 11% of breaches involved social engineering[52]
Verified
66NIST SP 800-53 Rev. 5 contains 20 families of security controls[53]
Single source
67NIST SP 800-53 Rev. 5 has 20 control families[53]
Directional
Cloud Security, Reliability & Compliance Interpretation
Cloud computing in 2026 is powered by mature reliability and security claims, like major providers backing multi-nines uptime, GDPR and HIPAA setting real financial stakes, and industry research repeatedly pointing to the same villains such as stolen credentials and authentication-focused attacks, all while compliance frameworks quietly organize the chaos into something auditors can actually grade.
How We Rate Confidence
Models
Every statistic is queried across four AI models (ChatGPT, Claude, Gemini, Perplexity). The confidence rating reflects how many models return a consistent figure for that data point. Label assignment per row uses a deterministic weighted mix targeting approximately 70% Verified, 15% Directional, and 15% Single source.
Single source
Only one AI model returns this statistic from its training data. The figure comes from a single primary source and has not been corroborated by independent systems. Use with caution; cross-reference before citing.
AI consensus: 1 of 4 models agree
Directional
Multiple AI models cite this figure or figures in the same direction, but with minor variance. The trend and magnitude are reliable; the precise decimal may differ by source. Suitable for directional analysis.
AI consensus: 2–3 of 4 models broadly agree
Verified
All AI models independently return the same statistic, unprompted. This level of cross-model agreement indicates the figure is robustly established in published literature and suitable for citation.
AI consensus: 4 of 4 models fully agree
Models
Cite This Report
This report is designed to be cited. We maintain stable URLs and versioned verification dates. Copy the format appropriate for your publication below.
APA
Stefan Wendt. (2026, February 13). Ai In The Cloud Computing Industry Statistics. Gitnux. https://gitnux.org/ai-in-the-cloud-computing-industry-statistics
MLA
Stefan Wendt. "Ai In The Cloud Computing Industry Statistics." Gitnux, 13 Feb 2026, https://gitnux.org/ai-in-the-cloud-computing-industry-statistics.
Chicago
Stefan Wendt. 2026. "Ai In The Cloud Computing Industry Statistics." Gitnux. https://gitnux.org/ai-in-the-cloud-computing-industry-statistics.
References
- 1gartner.com/en/newsroom/press-releases/2024-01-25-gartner-says-worldwide-end-user-spending-on-cloud-services-to-total-679-billion-in-2024
- 2gartner.com/en/newsroom/press-releases/2024-02-20-gartner-forecasts-worldwide-cloud-infrastructure-services-spending-will-total-295-2-billion-in-2025
- 3gartner.com/en/newsroom/press-releases/2023-11-09-gartner-says-worldwide-end-user-spending-on-cloud-services-to-reach-710-billion-in-2024
- 4gartner.com/en/newsroom/press-releases/2024-05-10-gartner-says-worldwide-end-user-spending-on-cloud-services-to-reach-1-trillion-in-2027
- 5idc.com/getdoc.jsp?containerId=US51703524
- 7idc.com/getdoc.jsp?containerId=US51356524
- 6ibm.com/downloads/cas/ZGZ4J8QZ
- 38ibm.com/reports/data-breach
- 8mckinsey.com/capabilities/quantumblack/our-insights/the-economic-potential-of-generative-ai-the-next-productivity-frontier
- 9mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai-in-2023
- 10ieeexplore.ieee.org/document/10130617
- 11nvidia.com/en-us/data-center/resources/state-of-ai-in-enterprise/
- 18nvidia.com/en-us/data-center/a100/
- 19nvidia.com/en-us/data-center/h100/
- 20nvidia.com/en-us/data-center/l40s/
- 12d1.awsstatic.com/whitepapers/aws-state-of-cloud-report.pdf
- 13cloud.google.com/static/solutions/ai/cloud-ai-study.pdf
- 23cloud.google.com/tpu/docs/tpus
- 30cloud.google.com/compute/sla
- 43cloud.google.com/storage/sla
- 44cloud.google.com/bigquery/docs/sla
- 14arxiv.org/abs/2005.14165
- 15arxiv.org/abs/1810.04805
- 16arxiv.org/abs/1512.03385
- 17arxiv.org/abs/1706.03762
- 27arxiv.org/abs/2303.08774
aws.amazon.com- 21aws.amazon.com/ec2/instance-types/trn2/
- 22aws.amazon.com/ec2/instance-types/in2/
- 25aws.amazon.com/ec2/nitro/
- 26aws.amazon.com/ec2/graviton/
- 29aws.amazon.com/ec2/sla/
- 31aws.amazon.com/s3/sla/
- 32aws.amazon.com/rds/sla/
- 33aws.amazon.com/sagemaker/sla/
- 42aws.amazon.com/s3/storage-classes/
- 24learn.microsoft.com/en-us/azure/virtual-machines/workloads/ai/azure-ai-architecture/nd-h100-v5
- 28learn.microsoft.com/en-us/azure/aks/sla
docs.aws.amazon.com- 34docs.aws.amazon.com/awscloudtrail/latest/userguide/cloudtrail-intro.html
- 46docs.aws.amazon.com/kms/latest/developerguide/deleting-keys.html
- 47docs.aws.amazon.com/kms/latest/developerguide/key-administration.html
- 35nist.gov/ai/ai-risk-management-framework
- 36cisecurity.org/benchmark/kubernetes
- 37owasp.org/Top10/
- 50owasp.org/www-project-application-security-verification-standard/
- 39gdpr.eu/fines/
- 40hhs.gov/hipaa/for-professionals/compliance-enforcement/data/index.html
- 41ftc.gov/legal-library/browse/cases-proceedings/penalties
- 45azure.microsoft.com/en-us/support/legal/sla/
- 48cloudflare.com/learning/bots/what-is-robot-traffic/
- 51cloudflare.com/learning/security/ransomware-statistics/
akamai.com- 49akamai.com/blog/security-research/api-attacks-increased-in-2023/
- 52verizon.com/business/resources/reports/dbir/
- 53csrc.nist.gov/pubs/sp/800/53/rev-5/final