Public cloud spending is forecast to grow by over 20% this year to approach $679 billion. This massive scale is enabling a parallel surge in AI investment, which analysts project will reach $300 billion in the same period.
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
Public cloud spend is set to surge from $563B in 2023 to $832.1B in 2025.
Related reading
01 · Category
Cloud Market & Spending20 stats
01
In 2023, cloud computing represented 10% of total global IT spending ($536 billion out of $5.53 trillion)
02
Gartner forecast worldwide end-user spending on public cloud services to total $679.0 billion in 2024
03
Gartner forecast worldwide end-user spending on public cloud services to total $832.1 billion in 2025
04
Gartner forecast worldwide end-user spending on cloud services to grow 20.4% in 2024 to $679.0 billion
05
Gartner forecast worldwide end-user spending on cloud services to grow 22.1% in 2025 to $832.1 billion
06
Gartner forecast global end-user spending on cloud infrastructure services to total $286.0 billion in 2024
07
Gartner forecast global end-user spending on cloud infrastructure services to total $295.2 billion in 2025
08
Gartner forecast global end-user spending on cloud infrastructure services to grow 19.0% in 2024
09
Gartner forecast global end-user spending on cloud infrastructure services to grow 13.1% in 2025
10
Gartner forecast worldwide end-user spending on public cloud services to grow 20.4% in 2024
11
In 2023, cloud services revenues were $563.0 billion worldwide
12
The right-to-left “cloud market” total for 2023 included SaaS, IaaS, PaaS, etc., and Gartner estimated $563.0 billion for 2023 cloud services
13
Gartner forecast worldwide cloud services spending to total $710.0 billion in 2024
14
Gartner forecast worldwide end-user spending on cloud services to reach $710.0 billion in 2024 (earlier forecast)
15
Gartner forecast worldwide end-user spending on cloud services to reach $1.0 trillion by 2027
16
Gartner forecast worldwide end-user spending on cloud services to reach $1.0 trillion by 2027
17
IDC forecast worldwide public cloud spending to reach $679.0 billion in 2023
18
IDC forecast worldwide public cloud spending to reach $1.1 trillion in 2027
19
IDC forecast public cloud spending to grow at a 16.9% CAGR from 2023 to 2027
20
IBM reported that the cost of AI chips and systems is being reduced by economies of scale in cloud usage (context)
Interpretation
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.
02 · Category
AI Demand & Adoption30 stats
01
IDC forecast global AI spending to reach $300+ billion by 2024 (AI spending market)
02
IDC forecast global AI spending to reach $300.8 billion in 2024
03
IDC forecast global AI spending to grow from $209.1 billion in 2023 to $300.8 billion in 2024
04
IDC forecast global AI spending will reach $500.0+ billion by 2026 (AI spending growth)
05
IDC forecast AI spending to reach $554.0 billion in 2025
06
McKinsey reported that generative AI can add $2.6to $4.4 trillion annually to business value
07
McKinsey estimated generative AI could create $200 billion to $340 billion in value for marketing and sales functions
08
McKinsey estimated generative AI could create $140 billion to $180 billion in value for customer operations
09
McKinsey estimated generative AI could create $60 billion to $110 billion in value for software engineering
10
McKinsey estimated generative AI could create $110 billion to $170 billion in value for IT functions
11
McKinsey estimated generative AI could create $90 billion to $150 billion in value for R&D
12
McKinsey reported that 63% of companies have adopted AI in at least one business function
13
McKinsey reported that 38% of companies have adopted AI in at least one business function in production or for internal use
14
McKinsey reported that AI adoption in the “production” stage increased by 8 percentage points from 2021 to 2022
15
McKinsey reported that 97% of surveyed companies have no single AI strategy owner
16
McKinsey reported that 37% of organizations are scaling AI
17
McKinsey reported that organizations that use AI well are 1.6x more likely to be top performers
18
IEEE reported that cloud computing is one of the most used platforms for AI deployment (contextual statistic)
19
NVIDIA reported that 2024 is expected to be a year of accelerated AI adoption in enterprises (forecast)
20
NVIDIA reported that 86% of enterprises are planning to deploy generative AI in the next 12-18 months (survey-based)
21
NVIDIA reported that 63% of enterprises plan to use generative AI for customer-facing applications (survey-based)
22
NVIDIA reported that 71% of enterprises are prioritizing accelerating model development and training (survey-based)
23
NVIDIA reported that 57% of enterprises are using or planning to use NVIDIA accelerated computing for AI (survey-based)
24
AWS reported that 65% of organizations plan to use AI/ML for competitive advantage (survey)
25
AWS survey indicated 67% of organizations plan to use AI/ML to improve operational efficiency
26
AWS survey indicated 58% plan to use AI/ML to improve customer experience
27
AWS survey reported that 77% of respondents said data and analytics are important to their organization’s future (context for AI)
28
AWS survey reported that 71% of respondents said they have a cloud-first strategy (helps AI in cloud adoption)
29
Google Cloud and IDC: 64% of IT executives say they are using AI to improve decision-making (report-based)
30
Google Cloud and IDC: 52% say they are using AI to automate business processes (report-based)
Interpretation
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.
More related reading
03 · Category
AI Infrastructure & Performance30 stats
01
OpenAI estimated that training large language models can require large compute clusters; (quantitative) GPT-3 had 175B parameters
02
GPT-3 model had 175 billion parameters
03
GPT-3 training compute used an amount of floating point operations described in the paper (FLOPs)
04
GPT-3 uses transformer architecture; model size range included 175B as largest
05
BERT-base has 110M parameters
06
BERT-large has 340M parameters
07
ResNet-152 has 60.2 million parameters
08
Transformer model in “Attention Is All You Need” has 65.7 million parameters for the base configuration
09
“Attention Is All You Need” reports a learning rate schedule with warmup steps of 4000
10
“Attention Is All You Need” uses 8 attention heads for base transformer
11
“Attention Is All You Need” uses a model dimension d_model=512 for base
12
“Attention Is All You Need” uses dropout rate 0.1
13
A100 Tensor Core GPU provides up to 312 TFLOPS (FP16) and 624 TFLOPS (TF32) per NVIDIA spec
14
NVIDIA A100 up to 19.5 TFLOPS (FP64) per NVIDIA spec
15
NVIDIA H100 Tensor Core GPU provides up to 1.979 PFLOPS (FP8) per NVIDIA spec
16
NVIDIA H100 provides up to 3.96 PFLOPS TensorFloat-32 (TF32) per spec
17
NVIDIA H100 provides up to 60 TFLOPS (FP64) per spec
18
NVIDIA L40S provides up to 499 TFLOPS (FP16) per spec
19
NVIDIA L40S provides up to 1200 GB/s memory bandwidth per spec
20
AWS Trainium2 supports up to 2.4 PFLOPS (bf16) per NVIDIA/AWS spec (AWS Trainium2)
21
AWS Trainium2 instances provide up to 2,304,000,000,000 (2.3 exa?) operations? (instance spec states up to 2.4 PFLOPS bf16)
22
AWS Inferentia2 provides up to 750 TOPS for INT8 per spec
23
AWS Inferentia2 provides 1,000 TOPS for INT8 per spec? (use stated maximum)
24
Google TPU v5e provides up to 4,000 TFLOPS (BF16) per spec
25
Google TPU v5e provides up to 1.6 TB/s memory bandwidth (or interconnect spec)
26
Azure ND H100 v5 VM uses NVIDIA H100 Tensor Core GPUs
27
Azure ND H100 v5 supports GPU count (8 GPUs per VM configuration)
28
AWS Nitro system offloads virtualization tasks to hardware
29
AWS Graviton processors are designed to deliver up to 40% lower cost compared to comparable x86 instances (per AWS)
30
AWS says Graviton3 delivers up to 25% better price performance for some workloads
Interpretation
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.
04 · Category
Cloud Security, Reliability & Compliance30 stats
01
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)
02
Azure Kubernetes Service SLA: 99.95% availability
03
AWS EC2 SLA availability is 99.99% for regions
04
Google Cloud compute engine SLA: 99.5% (or higher) availability for a month (per SLA document)
05
AWS S3 SLA: 99.9% availability
06
AWS RDS SLA: 99.99% availability
07
AWS SageMaker SLA: 99.9% availability (varies by feature)
08
AWS CloudTrail delivers logs within 15 minutes of creation (near real-time)
09
AWS says CloudTrail logs are delivered to an S3 bucket within 5 minutes? (depends) — use documented statement: up to 15 minutes
10
Microsoft 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
11
NIST AI Risk Management Framework (AI RMF 1.0) is organized into 4 core functions
12
CIS Kubernetes Benchmark v1.5 includes 270 scored benchmarks (checks)
13
OWASP Top 10: 2021 includes 10 categories
14
OWASP Top 10 2021 lists 10 categories of web application security risks
15
SANS report: average cost of a data breach in 2023 was $4.45 million (IBM)
16
IBM Security report: average cost of a data breach in 2023 was $4.45 million
17
IBM Security report: average time to identify a breach in 2023 was 207 days
18
IBM Security report: average time to contain a breach in 2023 was 72 days
19
IBM Security report: 53% of breaches involved compromised credentials
20
IBM Security report: 17% of breaches involved cloud misconfigurations (or use of cloud)
21
IBM Security report: 69% of breaches took place on premises vs cloud? (unclear)
22
AWS 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)
23
Under GDPR, administrative fines can be up to €20 million or 4% of total worldwide annual turnover (whichever is higher)
24
Under GDPR, administrative fines for certain infringements can be up to €10 million or 2% of annual turnover
25
Under HIPAA, civil penalties range from $137to $68,923 per violation depending on willful neglect vs not (numeric)
26
HIPAA civil penalties minimum is $137per violation (adjusted for inflation)
27
HIPAA civil penalties maximum is $68,923per violation
28
FTC can impose penalties up to $46,517per violation (as of 2023)
29
FTC penalty amount is $46,517per violation (inflation-adjusted)
30
Use cloud performance metric: AWS says S3 offers 11 9s durability (11 nines)
Interpretation
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.
Reference
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.
Sources & references
41 datasets cited across this report · attribution is report-level
aws.amazon.comdocs.aws.amazon.com
akamai.com+18 additional datasets cited (not shown individually)