Is the Generative AI Market growing now?
In our generative AI market deck, you will find everything you need to understand the market
SUMMARY
Yes, the Generative AI market is growing now, and the strongest evidence comes from customer spending, paid usage, application revenue and enterprise deployment rather than from funding headlines alone.
Enterprise spending has moved from experimentation into real software budgets. Menlo Ventures estimates spending rose from $1.7 billion in 2023 to $37 billion in 2025, with applications already taking more than half of the market.
The market is broadening below the foundation-model layer. Coding, healthcare, legal and other workflow products are building substantial recurring-revenue businesses, which makes Generative AI look less like one chatbot category and more like a collection of software markets.
AI coding is the clearest early killer market. Enterprise spending on coding tools rose from roughly $550 million in 2024 to $4 billion in 2025, while products such as Cursor and Lovable kept growing sharply into 2026.
Foundation-model competition is getting tougher even as the overall market expands. Anthropic gained major enterprise share, OpenAI lost some, and Google climbed quickly, showing that provider share can change dramatically without slowing category growth.
Falling inference prices have not reduced demand so far. Cheaper models appear to be creating more usage by making longer contexts, more agent steps, generated code, video and background automation economical enough to run at scale.
Enterprise rollout is deeper than it was a year ago, but the market still has plenty of runway. McKinsey found 44% of organizations scaling AI across the enterprise, meaning most companies still have not rolled it out broadly.
The weak point is ROI. Personal productivity gains are widespread, but only 37% of surveyed organizations report any positive EBIT impact from AI and just 6% qualify as high performers, so spending is clearly running ahead of proven company-wide economics.
Funding remains enormous but is much less broad than the headline totals suggest. A handful of giant foundation-model rounds can dominate an entire quarter, which makes customer spending a cleaner measure of market health than venture financing.
Big Tech capex supports the growth case because it is arriving alongside rising cloud revenue, backlog and AI consumption. The build-out still carries overcapacity risk, but it is not happening in a vacuum.
The market therefore looks fast-growing rather than mature. Generative AI demand has spread across models, enterprise applications, consumer usage and specialized workflows, but competition, falling prices and unresolved ROI will decide how much of today’s growth turns into durable profit.
This market map, featured in our generative AI market deck, highlights top companies and startups in the generative AI market
What should we actually count as the Generative AI market?
For this analysis, the Generative AI market means customer spending on foundation models, AI-native applications and the software layer that makes those products work.
That boundary matters when the numbers get huge. Nvidia GPUs, new power plants and entire hyperscaler data centers benefit from Generative AI, but counting all of them as Generative AI revenue would make the market almost impossible to measure cleanly.
Menlo Ventures gives us one of the more useful estimates because it looks at what enterprises actually spend on model APIs, AI infrastructure and applications. Its latest study puts that market at $37 billion in 2025. Stanford's AI Index gives us a very different number: $170.9 billion of private investment went into Generative AI companies that year. Both figures are useful, but only the first is close to measuring customer demand.
We therefore use revenue, enterprise spending and paid usage as the strongest evidence of market growth. Funding and infrastructure spending can confirm the picture, but they cannot prove it on their own.
| Measure | Latest figure | What it tells us |
|---|---|---|
| Enterprise Generative AI spending | $37B | What companies are paying for models, infrastructure and applications |
| Private investment in Generative AI companies | $170.9B | How aggressively investors are financing the sector |
| Hyperscaler AI capex | Hundreds of billions | How much capacity suppliers expect the market to need |
If you want more recent data on this point, please see our latest generative AI market report.
Is enterprise Generative AI spending still growing fast?
Enterprise Generative AI spending is still growing at a speed that would look extreme in almost any other software market.
Menlo Ventures estimates that enterprise spending went from $1.7 billion in 2023 to $11.5 billion in 2024 and then $37 billion in 2025. The market expanded roughly 22-fold in two years, including another 3.2-fold increase in the latest full year.
The mix of spending has also changed. Applications captured $19 billion, more than half of the total. Horizontal products such as AI assistants reached $8.4 billion, departmental applications such as coding and sales tools reached $7.3 billion, and vertical applications reached $3.5 billion.
That makes the growth more convincing. Enterprises are increasingly paying for software used by employees rather than concentrating the entire budget on models and technical infrastructure.
| Year | Enterprise Generative AI spend | Annual change |
|---|---|---|
| 2023 | $1.7B | Baseline |
| 2024 | $11.5B | 6.8x |
| 2025 | $37.0B | 3.2x |
As this chart shows, and as featured in our generative AI market deck, search interest in LLMs has surged
Is Generative AI still growing right now, or did it peak last year?
Generative AI growth is still visible in the freshest company results, with cloud usage, AI revenue and paid seats continuing to climb.
Microsoft's latest fiscal-year results showed Azure and other cloud services growing 43% year over year. Microsoft 365 Copilot has now passed 30 million paid seats. One quarter earlier, Microsoft said its AI business had already crossed a $37 billion annual revenue run rate while growing 123%.
Google is moving even faster in cloud. Google Cloud revenue recently jumped 82% year over year to $24.8 billion, with Alphabet specifically pointing to AI infrastructure and enterprise AI solutions. Gemini APIs processed about 22 billion tokens per minute, versus 16 billion only one quarter earlier, a 37.5% increase.
Amazon gives us a third independent check. AWS grew 37%, its fastest growth in 18 quarters, while Amazon said its AI business had passed a $25 billion annual revenue run rate and was growing at a triple-digit percentage. Bedrock also added more customers during the last six months than during its first two years after launch, and customers spent more on Bedrock in the latest quarter than in all previous quarters combined.
Three hyperscalers with different products are seeing the same thing these days: companies are consuming more AI, not less.
Are companies actually rolling Generative AI out across the business?
Companies are moving Generative AI into real workflows now, although enterprise-wide rollout is still far from finished.
McKinsey's latest global AI survey, fielded across 97 countries in May and June 2026, found that 44% of organizations are scaling AI across the enterprise, up from 38% a year earlier. The share using AI in at least three business functions also moved from 51% to 56%.
Large companies are much further ahead. Among organizations with at least $1 billion in annual revenue, 54% are scaling AI across the enterprise, compared with roughly one-third of smaller companies.
We are therefore seeing a different type of growth from the initial ChatGPT boom. More companies are now deepening deployments they already started. Marketing teams, developers, support operations, knowledge workers and IT departments are gradually pulling AI into everyday work.
The remaining runway is still large. A 44% scaling rate means most organizations have yet to roll AI out broadly, even after adoption has become mainstream.
This chart, featured in our generative AI market deck, shows annual VC investment in generative AI startups
Are consumers still using Generative AI more?
Consumer Generative AI usage is still expanding, and the latest data shows people using these tools more deeply after they adopt them.
Stanford's 2026 AI Index estimates that Generative AI reached about 53% population-level adoption within three years, faster than either the personal computer or the internet reached comparable penetration. The same research estimates U.S. consumer surplus from Generative AI at $172 billion annually, up 54% in one year.
The individual products have reached internet-platform scale. OpenAI now says its models reach more than one billion active users. Google recently reported 950 million monthly active users for the Gemini app, roughly three times the level from a year earlier.
OpenAI's usage data adds something more useful than another user count. Six months after signing up, people send roughly 50% more messages per day and use ChatGPT for about twice as many types of work. Growth is therefore happening inside existing accounts as well as through new users.
A lot of that consumer usage remains free, so one billion users should never be confused with one billion paying customers. Still, an expanding user base that becomes more active over time gives Generative AI companies a much larger surface for subscriptions, workplace conversion, APIs, advertising and commerce.
Is the Generative AI market still basically OpenAI, Anthropic and Google?
The foundation-model market is still concentrated, but leadership is moving around fast enough that no provider owns the category.
Menlo Ventures estimates that Anthropic captured 40% of enterprise LLM spending in 2025, up from 12% two years earlier. OpenAI moved the other way, from 50% to 27%, while Google increased from 7% to 21%. Together, those three companies still accounted for 88% of enterprise LLM spending.
Model performance is also bunching together. Stanford's latest AI Index found Anthropic, xAI, Google and OpenAI within 25 Elo points of one another on the Arena leaderboard. A gap that once looked enormous can now change after one model release.
This creates an unusual market. Generative AI can keep growing rapidly even when OpenAI loses share, Anthropic gains it or Google catches up. Money is moving between providers inside an expanding category.
Competition should get more aggressive from here. Once several models are good enough for the same job, price, reliability, distribution and integration start deciding more purchases.
If you want more recent data on this point, please see our latest generative AI market report.
This chart, featured in our generative AI market deck, looks at OpenAI’s strategy in generative AI
Are Generative AI apps turning into real software businesses?
Generative AI applications are now producing revenue at the scale of major software companies.
Cursor reportedly reached about $4 billion in annualized revenue in June 2026, up from $2 billion only four months earlier. Lovable reached a $500 million annualized run rate after crossing $400 million earlier in the year. The company has since raised $400 million at a $13.3 billion valuation.
The growth is spreading beyond coding. AI video company Higgsfield reportedly reached a $500 million revenue run rate after sitting near $50 million nine months earlier. Harvey, which sells AI to law firms and professional-services companies, says it added more than $100 million of ARR during a single quarter.
Those companies sell four very different jobs: professional coding, software creation for non-developers, commercial video generation and legal work. Seeing hundreds of millions or billions of dollars appear across several workflows is much stronger evidence than one breakout chatbot.
The application layer is where Generative AI starts looking most like a normal software market: users have specific jobs to do, products compete for budgets and the winners can build substantial recurring revenue.
Is AI coding already Generative AI's first killer market?
AI coding is currently Generative AI's clearest killer market.
Menlo Ventures estimates enterprise spending on AI coding products reached $4 billion in 2025, up from roughly $550 million one year earlier. That is more than sevenfold growth, and coding alone captured around 55% of departmental AI spending.
Company growth has continued after that measurement period. Cursor moved from roughly $2 billion in annualized revenue in February 2026 to about $4 billion by June. Lovable reached $500 million while broadening AI software creation beyond professional developers.
Buyer behavior is changing too. McKinsey's latest survey found that 32% of respondents said their organizations had skipped at least one software product or feature because coding agents allowed them to build the functionality internally.
Generative AI spending in coding is therefore doing more than taking budget from traditional developer tools. It is beginning to change the build-versus-buy decision for software itself.
| AI coding evidence | Earlier level | Latest level |
|---|---|---|
| Enterprise AI coding spend | $550M in 2024 | $4.0B in 2025 |
| Cursor annualized revenue | ~$2B in Feb. 2026 | ~$4B in June 2026 |
| Lovable annualized revenue | ~$400M earlier in 2026 | $500M in June 2026 |
| Companies skipping software purchases because of coding agents | No comparable historical figure | 32% of McKinsey respondents |
If you want more recent data on this point, please see our latest generative AI market report.
This chart, featured in our generative AI market deck, shows annual funding in generative AI startups
Is vertical Generative AI becoming a real market?
Vertical Generative AI is becoming a real market, with healthcare and legal services pulling ahead.
Menlo Ventures estimates vertical AI spending reached $3.5 billion in 2025, up from $1.2 billion the previous year. Healthcare accounted for about $1.5 billion and legal AI roughly $650 million.
Healthcare now provides some unusually concrete evidence. Abridge says its platform is used by more than 300 health systems and processes more than 100 million clinical conversations a year. OpenEvidence was reportedly approaching a $300 million annualized revenue run rate by mid-2026, roughly double its level around the end of 2025.
Legal AI is developing along the same lines. Harvey said its business added more than $100 million of ARR during its latest quarter as law firms and professional-services organizations expanded usage.
The common thread is easy to see. Healthcare and legal businesses already spend enormous amounts on highly paid human work, documentation, research and administration. AI can go after those existing budgets directly. That gives vertical Generative AI a clearer economic foundation than markets built around novelty content or generic chat.
Are AI agents actually scaling inside companies?
AI agents are scaling now, especially inside large companies, but most businesses are still early in that transition.
McKinsey's latest survey found that 40% of respondents at companies with more than $1 billion in revenue were scaling AI agents in at least one function, up from 27% one year earlier. Smaller companies stayed almost flat at 22%.
Coding agents are moving particularly quickly. Around one-fifth of all organizations said they were scaling them, rising to 31% among large enterprises. IT, knowledge management and software engineering currently lead agent adoption.
The 32% build-versus-buy figure from McKinsey is especially revealing here. Agentic coding has already become capable enough for some companies to cancel planned software purchases and build features internally.
Still, enterprise-wide agent deployment remains far behind chatbots, which 47% of respondents say they are already scaling across the organization. Agents are contributing to Generative AI growth today, with much of their potential market still ahead.
This chart, featured in our generative AI market deck, compares the main business model options for generative AI SaaS platforms
Is Generative AI producing real ROI yet?
Generative AI is producing obvious productivity gains, but company-wide financial returns are still much weaker than adoption.
McKinsey's newest global survey found that 80% of respondents believe AI improves their personal productivity and about half say it helps them make better decisions. At the company level, only 37% attribute any positive EBIT impact to AI, almost unchanged from the previous year.
The gap gets wider when we look for substantial returns. Just 6% of respondents qualify as AI high performers, meaning they attribute at least 5% of EBIT to AI and describe its overall value as significant.
That weakness should be taken seriously. Generative AI spending has raced ahead of the financial results most companies can currently measure.
Companies are still behaving as though they expect the economics to improve. Sixty percent of respondents plan to increase AI investment during the coming year. Nearly three-quarters of the small group of high performers say they have fundamentally redesigned workflows around AI, compared with only one-quarter of the rest.
So the market is growing before most buyers have fully solved the ROI equation. That can continue for quite a while, but not forever.
If you want more recent data on this point, please see our latest generative AI market report.
Are falling AI prices hurting Generative AI revenue?
Falling AI prices are making each unit of intelligence cheaper, while usage is currently rising fast enough to keep the market growing.
The cost decline has been brutal. Stanford previously calculated that the cost of running a model at roughly GPT-3.5 performance fell from about $20 per million tokens in late 2022 to $0.07 by late 2024, more than a 280-fold drop.
Model competition keeps pushing prices lower, yet customers keep finding more things to run. Google's Gemini API volume recently increased from 16 billion to 22 billion tokens per minute in a single quarter. Amazon says companies are increasingly moving inference workloads into production, helping push its AI business above a $25 billion annual revenue run rate.
There is also a straightforward reason this can work economically. When inference becomes cheaper, developers can afford longer context windows, more agent steps, more generated code, more video and more automated background work. Lower prices can create workloads that were previously too expensive.
McKinsey does show the limit. Twenty percent of respondents say AI operating costs, including token costs, already constrain their usage. Cost still matters a lot.
For now, cheaper inference looks more like a demand accelerator than a market killer.
This chart, featured in our generative AI market deck, illustrates how revenue is distributed across customer segments in the generative AI market
Are foundation models becoming interchangeable?
Foundation models are becoming easier to substitute on many tasks, even though the best closed models still keep a small performance edge.
Stanford's latest AI Index shows how tight the frontier has become. Anthropic, xAI, Google and OpenAI sat within just 25 Elo points on the Arena leaderboard. The gap between the best closed and best open model was about 3.3% as of March 2026.
That is enough differentiation to matter on demanding workloads, but nowhere near the gap we saw during the first phase of the market. Companies can increasingly route simple work to cheaper models, reserve expensive models for hard tasks and change providers without rebuilding the whole product.
Enterprise spending already reflects that behavior. Anthropic gained enormous share while OpenAI lost share and Google climbed quickly, yet the overall model market kept expanding.
This should put pressure on model margins over time. The bigger opportunity may gradually move toward distribution, proprietary data, workflow integration and applications where switching means changing how people actually work.
Is Generative AI funding still accelerating?
Generative AI funding is still enormous, but the headline totals are now so concentrated that they exaggerate how broad the financing boom is.
S&P Global counted a record $217.7 billion raised by Generative AI companies during the first half of 2026, already about twice its comparable total for all of 2025. On the surface, funding appears to be accelerating even faster than the market.
The underlying numbers look very different. S&P recorded about $145 billion in the first quarter alone, and two rounds accounted for roughly 98% of it: $122 billion for OpenAI and $20 billion for xAI. That leaves only around $3 billion for everybody else in that quarter.
Funding then fell to roughly $72.7 billion in the following quarter. That is around half the previous quarter's total, although S&P still describes it as the second-strongest quarter in the history of its dataset.
Capital is geographically concentrated too. Stanford counted about $163.6 billion of U.S. private Generative AI investment out of $170.9 billion globally in 2025, or roughly 96%.
The funding boom is real, but these days it is heavily shaped by a handful of gigantic foundation-model companies. That makes funding a weaker measure of broad market health than customer spending.
| Funding measure | Amount | What we learn |
|---|---|---|
| Global private Generative AI investment in 2025, Stanford | $170.9B | Capital more than tripled from 2024 |
| S&P Generative AI fundraising in H1 2026 | $217.7B | Record amount of capital is still entering the sector |
| S&P Generative AI fundraising in Q1 2026 | ~$145B | One extraordinary quarter drove much of H1 |
| OpenAI + xAI share of Q1 funding | ~98% | Funding growth is extremely concentrated |
This chart, featured in our generative AI market deck, shows how AI video generation technology has evolved over time
Does Big Tech's AI spending prove real Generative AI demand?
Big Tech's AI spending strengthens the growth case only when we pair it with actual customer consumption, and the latest results show plenty of both.
The infrastructure bill is enormous. S&P Global's latest Visible Alpha monitor estimates that capital spending across major technology companies could reach roughly $700 billion in 2026. Alphabet alone recently raised its expected annual capex range to $195 billion to $205 billion.
Spending that much creates obvious overbuilding risk. What makes today's situation more convincing is the revenue sitting beside the construction.
Google Cloud recently grew 82% to $24.8 billion and ended the quarter with $514 billion of backlog. AWS grew 37% to $42.2 billion for the quarter and reported $496 billion of backlog. Microsoft's Azure grew 43%, while Microsoft's commercial remaining performance obligation reached $678 billion.
Those backlog figures include plenty of non-AI cloud business, so we should not label the full amounts Generative AI demand. The companies themselves, however, repeatedly identify AI infrastructure, AI solutions and inference workloads as major growth drivers.
There will almost certainly be individual data centers that earn disappointing returns. The broader build-out currently has enough revenue growth behind it to look like capacity being pulled forward for a real market rather than infrastructure built for imaginary customers.
If you want more recent data on this point, please see our latest generative AI market report.
So, is the Generative AI market growing now?
Yes, the Generative AI market is growing now, and the latest evidence makes that a strong conclusion rather than a forecast.
As seen above, Menlo's multi-year enterprise spending series is still moving sharply upward, with applications already taking more than half of measured spending. More recent data shows that the momentum continued afterward: Azure grew 43%, Google Cloud grew 82%, AWS grew 37%, and Amazon's AI business is growing at triple-digit rates.
Usage tells the same story from another angle. ChatGPT now reaches more than one billion active users, Gemini is approaching the same scale, Google model API traffic increased by more than a third in one quarter, and OpenAI says existing users become more active as they spend more time with the product.
The strongest change is happening below the chatbot layer. Cursor reached roughly $4 billion in annualized revenue. Lovable reached $500 million. Harvey added more than $100 million of ARR in one quarter. OpenEvidence is approaching a $300 million run rate. AI coding, healthcare and legal software are turning Generative AI into separate commercial markets rather than one giant assistant category.
There is one serious weakness in the story: financial returns inside ordinary enterprises remain underwhelming. McKinsey finds that 44% of organizations are now scaling AI across the enterprise, yet only 37% report any positive EBIT impact and just 6% qualify as high performers. Companies are spending ahead of proven company-wide ROI.
That weakness keeps us from calling Generative AI a mature market. It does not make the current growth ambiguous.
The Generative AI market is growing quickly today across spending, usage, deployment and application revenue. Funding is more fragile and concentrated than the headlines suggest, model prices are falling, and many buyers still need to prove the economics. But the underlying demand has broadened enough that the market no longer depends on investors financing foundation models or consumers experimenting with chatbots. Companies are now paying for Generative AI to do actual work, and that commercial layer is still getting bigger.
In our generative AI market deck, we identify pain points entrepreneurs should prioritize
OUR METHODOLOGY
This analysis tests whether the Generative AI market is growing now by separating the category into customer spending, usage, enterprise deployment, application revenue, model competition, vertical AI, agents, pricing, funding, infrastructure demand and realized ROI.
We give the most weight to evidence closest to real demand: enterprise spending, paid seats, revenue, usage, deployments and consumption. Funding rounds and infrastructure capex are useful supporting evidence, but they are not treated as proof of market growth on their own.
Menlo Ventures provides the main multi-year spending baseline because its enterprise Generative AI work measures model APIs, AI infrastructure and applications rather than simply adding up investment into AI companies. Stanford HAI is used for private investment, consumer adoption, consumer surplus, technical-performance convergence and the decline in inference costs.
Recent company disclosures are used to test whether the broader trend remained visible after the latest full-year market datasets. Microsoft, Alphabet and Amazon provide cloud growth, AI revenue, paid-seat, token-volume, backlog and Bedrock consumption data. OpenAI provides current user-scale and engagement data.
McKinsey's 2026 State of AI survey is the main source for enterprise-wide deployment, agent adoption, productivity, EBIT impact, high performers and cost constraints. We use these figures to distinguish rapid adoption from proven financial returns inside ordinary companies.
Application-level evidence comes from companies operating in different workflows rather than one product category. Abridge and OpenEvidence are used for healthcare, Harvey for legal AI, Cursor and Lovable for coding and software creation, and Higgsfield for AI video. The point is to test whether meaningful commercial businesses are appearing across several jobs.
Funding is treated cautiously because the latest totals are heavily concentrated. S&P Global Market Intelligence is used for H1 2026 fundraising and the unusually large contribution from OpenAI and xAI, while Stanford HAI provides the broader 2025 private-investment baseline.
Key sources used for this analysis include Menlo Ventures' 2025 State of Generative AI in the Enterprise, Stanford HAI's 2026 AI Index economy chapter, Stanford HAI's technical-performance chapter, Stanford HAI's 2025 research and development chapter, Microsoft FY2026 Q3 earnings, Microsoft FY2026 Q4 results, Alphabet's Q2 2026 earnings remarks, Alphabet's Q2 2026 SEC filing, Amazon's Q2 2026 results, McKinsey's State of AI survey, OpenAI's ChatGPT adoption analysis, OpenAI's user-scale disclosure, Abridge's deployment disclosure, Harvey's strategic investment announcement, Lovable's Series C announcement, and S&P Global Market Intelligence on H1 2026 Generative AI fundraising.
This chart, featured in our generative AI market deck, illustrates how revenue is distributed across Europe, Asia, North America, Africa, and South America in the generative AI market
Related blog posts
- A complete list of business models in generative AI
