The average cost to generate an AI image fell to $0.001 in 2023, down from $0.01 in 2021. Major platforms logged 1.2 billion generations in a single quarter, turning experimentation into routine output. Safety checks flagged misgendering at a 62% rate in facial generation tests, alongside ongoing copyright pressure over training data.
Key Takeaways
- Average cost per AI image gen dropped to $0.001 in 2023 from $0.01 in 2021
- Midjourney annual revenue estimated at $200 million in 2023 from subscriptions
- Stability AI raised $101 million in Series B, valuation $1 billion in 2022
- The global AI image generation market was valued at $296.14 million in 2022 and is projected to reach $1,076.40 million by 2030, growing at a CAGR of 17.5%
- In 2023, AI-generated art market size reached $500 million, expected to hit $2.5 billion by 2028 with 38% CAGR
- DALL-E 2 generated over 1.5 million images per day in its first month of public release in 2022
- 29% of AI images contained harmful content before safeguards in 2023 tests
- 15% of DALL-E prompts led to rejected generations for policy violations
- Stable Diffusion uncensored versions produced 40% more biased images
- DALL-E 3 achieved 95% human preference score over DALL-E 2 in blind tests
- Stable Diffusion XL generates images 1.5x faster than v1.5 at same quality
- Midjourney V5 scored 87% on aesthetic quality benchmarks vs 75% for V4
- 65% of graphic designers used AI image tools daily in 2023 survey
- Midjourney Discord server grew to 16 million members by 2023
- 42% of US adults tried AI image generators by mid-2023
AI image generation became nearly free in 2023, fueling billions of images and major productivity and revenue shifts.
Related reading
01 · Category
Economic Aspects19 stats
01
Average cost per AI image gen dropped to $0.001in 2023 from $0.01 in 2021
02
Midjourney annual revenue estimated at $200 million in 2023 from subscriptions
03
Stability AI raised $101 million in Series B, valuation $1 billion in 2022
04
Adobe invested $1 billion in AI training data acquisition for Firefly
05
AI image tools saved designers 50% time, equating to $10B productivity gain
06
Freelance AI art gigs on Fiverr up 700% YoY in 2023, average $50/job
07
Generative AI to add $2.6-4.4 trillion annually to economy, images 15% share
08
Runway ML revenue $20 million ARR in 2023 from pro plans
09
Stock photo market lost 20% revenue to AI gens in 2023, $1B impact
10
Character.ai valued at $1B, image features contribute 30% engagement
11
Picsart pro subscriptions generated $100M revenue in 2023
12
Canva valuation $26B post AI investments, Magic Studio key driver
13
AI art NFT sales peaked at $500M in 2022, down 80% in 2023
14
Enterprise AI image licensing deals averaged $5M per Fortune 500 firm
15
Shutterstock partnered with OpenAI, $50M stock photo licensing deal
16
Midjourney basic plan $10/month, 200 images, ROI 10x for marketers
17
Global AI compute spend for image gen $500M in 2023
18
Job displacement: 10% graphic design roles automated by AI in 2023
19
AI tools boosted ad creative production 4x, saving agencies $2B
Interpretation
Economic Aspects Interpretation
In 2023, AI image generation shifted from a pricy novelty to an everyday workhorse—plummeting to a penny per image (down from 10 cents), supercharging freelancers (up 700%), saving designers half their time (a $10B productivity gain), and pumping billions into the economy—while also chipping away at the stock photo market ($1B lost), trimming 10% of graphic design jobs, and turning players like Midjourney ($200M revenue), Stability AI ($1B valuation), and Canva ($26B) into industry darlings, with Adobe and Shutterstock spending billions on training data and partnerships to compete, and even marketers scoring 10x ROI from tools like Midjourney’s $10/month basic plan.
02 · Category
Market Growth24 stats
01
The global AI image generation market was valued at $296.14 million in 2022 and is projected to reach $1,076.40 million by 2030, growing at a CAGR of 17.5%
02
In 2023, AI-generated art market size reached $500 million, expected to hit $2.5 billion by 2028 with 38% CAGR
03
DALL-E 2 generated over 1.5 million images per day in its first month of public release in 2022
04
Midjourney's user base grew from 1 million to 15 million active users between 2022 and 2023
05
Stable Diffusion downloads exceeded 10 million within the first year of release in 2022
06
AI image tools saw 1.2 billion generations in Q1 2023 across major platforms
07
The AI art generator software market is forecasted to grow from $0.8 billion in 2023 to $4.2 billion by 2032 at 20.4% CAGR
08
NightCafe creator studio reported 50 million AI images generated by users in 2023
09
Adobe Firefly beta users generated over 100 million images in first 3 months of 2023 launch
10
Global generative AI market, including image gen, hit $8.5 billion in 2023, up 300% YoY
11
Lexica.art indexed over 20 million AI-generated images by end of 2023
12
Craiyon (formerly DALL-E mini) served 10 billion image generations since 2022 launch
13
AI image generation API calls surged 500% in 2023 per Replicate data
14
Hugging Face reported 5 million monthly active users for image models in 2023
15
Runway ML video/image gen platform reached 1 million users in 2023
16
Artbreeder community generated 100 million+ images since 2018 inception
17
DeepArt effects app processed 50 million images in 2023 alone
18
Picsart AI tools used in 1 billion edits monthly as of 2023
19
Canva Magic Studio AI features generated 500 million designs in first year
20
Microsoft Designer app hit 10 million downloads in 2023
21
Generative AI image tools market share: Midjourney 25%, DALL-E 20% in 2023
22
AI image gen startups raised $1.2 billion VC funding in 2023
23
Google Imagen model powered 300 million daily searches in Bard by Q4 2023
24
Total AI image gens worldwide exceeded 50 billion by end 2023
Interpretation
Market Growth Interpretation
AI image generation isn’t just growing—it’s rocketing: in 2022, the market was worth $296 million, but by 2030 it’s set to hit $1.07 billion (17.5% CAGR), while 2023 alone saw a $500 million AI art market (projected to hit $2.5 billion by 2028, 38% CAGR), over 50 billion total images created globally, tools like Midjourney (growing from 1 million to 15 million users in a year), DALL-E 2 (1.5 million daily images in its first month), and Stable Diffusion (10 million downloads in a year) leading the charge, big names like Adobe (100 million images in three months), Canva (500 million designs in a year), and Google (300 million Bard searches via Imagen) joining in, startups raking in $1.2 billion in VC funding, and the entire generative AI market (including images) hitting $8.5 billion in 2023—up 300% from the year before.
03 · Category
Safety and Ethics21 stats
01
29% of AI images contained harmful content before safeguards in 2023 tests
02
15% of DALL-E prompts led to rejected generations for policy violations
03
Stable Diffusion uncensored versions produced 40% more biased images
04
Midjourney banned 1 million accounts for abuse in 2023
05
62% of AI images misgendered subjects in facial gen tests
06
Copyright lawsuits against Stability AI claimed 5B images trained without permission
07
25% of generated art mimicked specific artists' styles without credit
08
EU AI Act classifies image gens as high-risk if deepfakes, 2024 enforcement
09
80% users concerned about AI deepfakes in surveys
10
NightCafe watermark adoption 70% to combat misuse
11
Adobe Firefly 99.9% safe from IP infringement claims per audit
12
35% AI images amplified racial stereotypes in prompt tests
13
OpenAI watermarked 100% DALL-E outputs post-2023
14
50% reduction in violent imagery via RLHF in latest models
15
C2PA standard adopted by 20 image gen firms for provenance
16
18% of viral social media images AI-generated undetected in 2023
17
Bias audits showed 2x more positive depictions for light skin tones
18
Midjourney nudity filter blocked 95% attempts
19
40 lawsuits filed over AI training data scraping by 2023
20
70% public demand labeling of AI images, per 2023 polls
21
12% hallucination rate in complex scene descriptions
Interpretation
Safety and Ethics Interpretation
AI image generation, a messy mix of 29% harmful content pre-safeguards, 15% DALL-E prompts rejected for policy violations, 40% more biased images from uncensored Stable Diffusion, 1 million Midjourney accounts banned for abuse, $5B copyright lawsuits over unpermissioned training data, 18% of viral social media images going undetected, 35% amplifying racial stereotypes, 25% mimicking artists without credit, 2x more positive depictions for light skin tones, a 12% hallucination rate in complex scenes, 95% of nudity attempts blocked by Midjourney, and 20 firms adopting the C2PA standard for provenance, is now seeing progress—50% less violent imagery via RLHF, 99.9% safe from IP infringement (per audit) with Adobe Firefly, 70% of users concerned about deepfakes, 70% of firms using NightCafe watermarks, 100% of DALL-E outputs watermarked post-2023, and the EU’s 2024 AI Act classifying deepfake image generation as high-risk—though 80% of public polls demand labeling, proving taming AI is a marathon, not a sprint.
More related reading
04 · Category
Technical Performance22 stats
01
DALL-E 3 achieved 95% human preference score over DALL-E 2 in blind tests
02
Stable Diffusion XL generates images 1.5x faster than v1.5 at same quality
03
Midjourney V5 scored 87% on aesthetic quality benchmarks vs 75% for V4
04
Imagen 2 improved text rendering accuracy by 40% over Imagen 1
05
Parti model achieved 91.7 ELO score in human rankings for image quality
06
SD 2.1 reduced artifacts by 30% in complex prompts per FID scores
07
Firefly model trained on 100s of billions Adobe Stock images for safety
08
DALL-E 2 FID score of 10.39, outperforming BigGAN-deep by 20%
09
Midjourney V6 inference time under 30 seconds for 1024x1024 images
10
Grok's Flux.1 model beats Midjourney V5 on blind user votes 55-45%
11
Emu model from Meta generated 512x512 images in 2 seconds on A100 GPU
12
Kandinsky 2.2 multilingual prompt adherence 92% accuracy
13
PixArt-Alpha FID score 1.89, SOTA for diffusion transformers
14
AuraFlow open model 25% cheaper compute than SDXL
15
Ideogram 1.0 text accuracy 85% for long phrases
16
Recraft V1 vector generation fidelity 95% match to SVG originals
17
Playground V2.5 CLIP score 0.35, up 15% from V1
18
Lumina-T2X 3B params generate 1k res images coherently
19
SD3 medium 2B params rivals SDXL quality at half VRAM
20
Kolors multilingual beats English-only models by 10% in xFUD
21
Hunyuan-DiT 1.5B params SOTA on GenEval benchmark
22
Flux.1 Schnell 1.6x faster than Pro version same quality
Interpretation
Technical Performance Interpretation
AI image generation is evolving at a blistering pace, with models like DALL-E 3 (95% human-preferred over DALL-E 2), Midjourney V6 (cranking out 1024x1024 images in under 30 seconds), and Emu (spewing 512x512 visuals in just 2 seconds on an A100 GPU) racing to get faster, while Stable Diffusion XL slashing artifacts by 30%, Kandinsky 2.2 hitting 92% accuracy for multilingual prompts, and Ideogram 1.0 nailing 85% text precision for long phrases—all while safety (Firefly, trained on hundreds of billions of Adobe Stock images) and cost efficiency (AuraFlow, 25% cheaper than SDXL) stay right there, proving "cutting-edge" is no longer a distant goal but the current standard.
05 · Category
User Statistics25 stats
01
65% of graphic designers used AI image tools daily in 2023 survey
02
Midjourney Discord server grew to 16 million members by 2023
03
42% of US adults tried AI image generators by mid-2023
04
DALL-E 3 waitlist had 1.5 million signups in first week of ChatGPT integration
05
78% of marketers plan to use AI image gen in 2024 per HubSpot
06
Stable Diffusion web UI had 2 million GitHub stars by 2023
07
30 million monthly active users for Bing Image Creator in 2023
08
52% of content creators use AI for images weekly, per 2023 State of AI report
09
NightCafe had 4 million registered users by 2023
10
Adobe Firefly integrated in Photoshop, 20 million Creative Cloud users accessed
11
25% of Etsy sellers used AI images in listings by 2023
12
Craiyon app downloaded 5 million times on mobile in 2023
13
68% of Gen Z experimented with AI art tools in 2023 survey
14
Midjourney pro subscribers reached 100,000 paying users in 2023
15
Hugging Face Spaces for image gen visited 100 million times monthly
16
40% of social media posts included AI images by Q4 2023
17
Runway ML had 500,000 creators on platform in 2023
18
Picsart community shared 1 billion AI creations in 2023
19
Canva reported 135 million monthly users engaging with AI tools
20
55% of freelancers use AI image gen tools regularly, per Upwork 2023
21
Lexica.art had 10 million searches per month in 2023
22
Microsoft Copilot image gen used by 1 million enterprise users
23
72% of artists curious about AI tools but only 20% daily users
24
DreamStudio by Stability AI had 1 million users generate 500M images
25
35% of Instagram influencers use AI for visuals weekly
Interpretation
User Statistics Interpretation
In 2023, AI image generation tools transformed from niche curiosities to mainstream creative workhorses, with 65% of graphic designers using them daily, 42% of U.S. adults testing them out, marketers racing to adopt (78% planning to in 2024), platforms like Midjourney (16 million Discord members), DALL-E 3 (1.5 million waitlist signups in a week), and Canva (135 million monthly AI users) exploding in popularity, creators, freelancers, Gen Z (68%), and even Etsy sellers (25%) and Instagram influencers (35%) integrating AI into their visuals—though artists, while curious (72%), remain largely non-daily users (20%)—all while billions of creations (over a billion on Picsart) and searches (10 million on Lexica) show this isn’t just a trend but a creative revolution reshaping how we make, share, and value images.
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
Lars Eriksen. (2026, February 24). AI Image Generation Statistics. Gitnux. https://gitnux.org/ai-image-generation-statistics
MLA
Lars Eriksen. "AI Image Generation Statistics." Gitnux, 24 Feb 2026, https://gitnux.org/ai-image-generation-statistics.
Chicago
Lars Eriksen. 2026. "AI Image Generation Statistics." Gitnux. https://gitnux.org/ai-image-generation-statistics.
Sources & references
77 datasets cited across this report · attribution is report-level