How big is the generative AI market really?
Whether you're building or investing in the generative AI market, our report covers everything you need to know: the players, the money, and what's coming next.
SUMMARY
The direct generative AI market is already in the low hundreds of billions of dollars, while the wider AI economy being built around it is measured in the trillions.
The apparent gap between a $40 billion market and a $2.7 trillion one is mostly a definition problem. Consumer subscriptions, enterprise applications, foundation models, cloud infrastructure, consulting, chips and data centers are all real AI businesses, but they are not the same market.
The strongest floor comes from observable company revenue. OpenAI and Anthropic alone recently exceeded $105 billion of combined annualized revenue, which makes older estimates of a direct GenAI market worth only a few tens of billions increasingly hard to defend.
Consumer and enterprise spending are both becoming meaningful at the same time. Menlo Ventures estimates roughly $40 billion of consumer AI spending and $37 billion of enterprise GenAI spending, with the consumer jump driven more by deeper monetization than by explosive user growth.
Enterprise spending has already moved up the stack. Applications captured about $19 billion in Menlo’s 2025 estimate, more than foundation models, model training and AI infrastructure software combined.
AI coding is the clearest proof that a standalone application category can become large quickly. It already represents billions of dollars of enterprise spending, while coding agents can generate very high model usage because one task may trigger many calls, tests and retries.
Model revenue remains concentrated, but customer behavior is much less exclusive. OpenAI, Anthropic and Google dominate enterprise API spending, while both consumers and companies increasingly use several assistants or models at once.
Double counting is the biggest trap in market sizing. One end-customer payment can appear again as application revenue, model-provider revenue, cloud revenue and ultimately hardware revenue, so adding disclosures across the supply chain can wildly overstate final demand.
Infrastructure is running far ahead of direct AI revenue. The largest hyperscalers are collectively planning more than $700 billion of annual capital expenditure, which means the industry is building for a workload several times larger than the one customers pay for today.
Falling inference prices do not automatically shrink the market. Cheaper models are making longer-running agents and more automated tasks economical, so the relevant unit is shifting from the price of a token toward the cost of completing useful work.
Announcements are easy, delivery is harder. Our generative AI market report checks what actually got built.
How can the generative AI market be worth $40 billion and $2.7 trillion at the same time?
The huge gap between generative AI market estimates comes mostly from what researchers decide to count, and the numbers become much less contradictory once we separate AI products from the infrastructure built around them.
Take Gartner’s latest forecast. It puts worldwide AI spending at roughly $2.67 trillion in 2026. AI infrastructure alone accounts for about $1.48 trillion. AI services add $576 billion and AI software another $462 billion. By comparison, Gartner expects spending specifically on generative AI models to reach about $28 billion, with AI agents and assistants adding around $29 billion.
Those numbers all describe genuine economic activity, but they answer very different questions.
A company paying for Claude or ChatGPT is buying a generative AI product. A hyperscaler buying thousands of accelerators is building capacity that could support many AI products for years. A business paying consultants to connect AI to its internal systems creates another layer of revenue. Counting the full value of all three can be useful when measuring the economic ecosystem around AI, but it produces a very inflated answer if our question is simply how much customers currently spend on generative AI itself.
That distinction explains most of the confusion around the market.
For this article, we treat money paid for generative models, AI applications and AI-native services as the core generative AI market. Hardware, cloud capacity, data centers and supporting services remain essential to the analysis because they show what the industry expects to happen next, but we keep them separate from direct demand wherever possible.
| What is being measured? | Recent scale | What the number really tells us |
|---|---|---|
| Consumer AI products | ~$40B | What consumers are estimated to spend on AI products |
| Enterprise GenAI | $37B in 2025 | What companies spent on models, AI infrastructure software and applications |
| Generative AI models | ~$28B forecast | Direct spending on model providers |
| AI agents and assistants | ~$29B forecast | Spending on agentic and assistant products |
| All AI software | ~$462B forecast | Much broader than generative AI |
| AI services | ~$576B forecast | Consulting, implementation and related work |
| AI infrastructure | ~$1.48T forecast | Chips, servers, networks, cloud capacity and devices |
| All AI spending | ~$2.67T forecast | The full AI economy under Gartner’s definition |
If you want more recent data on this point, please see our latest generative AI market report.
How much money are customers actually paying for generative AI today?
Direct generative AI revenue is already pushing into the hundreds-of-billions range, although the available data still does not support one clean global total.
The clearest floor comes from the companies selling the models.
OpenAI’s annualized revenue reportedly passed $40 billion recently, roughly double its pace at the end of 2025. Anthropic’s annualized revenue passed $65 billion by the end of July, according to figures reported by Reuters and Bloomberg. Anthropic had been around $47 billion only two months earlier and roughly $9 billion at the end of 2025.
Those two companies alone were therefore running above $105 billion of combined annualized revenue at their latest independently reported levels.
We need to be careful with that number. A revenue run rate extrapolates a recent sales pace over a full year, so $105 billion is not the amount OpenAI and Anthropic actually collected during the previous twelve months. The companies may also calculate their run rates differently.
Still, the order of magnitude is hard to ignore.
And OpenAI plus Anthropic represents only part of the market. Google sells Gemini through consumer subscriptions, enterprise products and APIs. Microsoft sells Copilot across productivity, coding and security. Amazon now says its AWS AI business alone has passed a $25 billion annual revenue run rate. Then there are coding products, image and video generators, legal AI, healthcare AI, customer-support platforms and hundreds of smaller applications.
Trying to add all of those businesses together would create plenty of double counting, especially when an application vendor pays a model provider that then pays a cloud provider. But the evidence is already strong enough to reject the idea that generative AI remains a market worth merely a few tens of billions of dollars.
Speed of adoption says a lot about a market's potential. See how fast things are moving in our generative AI market report.
How big is the consumer generative AI market now?
Consumer AI is currently about a $40 billion global market by Menlo Ventures’ latest estimate, and most of the recent growth came from existing users spending more rather than millions of new people suddenly discovering AI.
Menlo’s 2026 consumer study is useful here because it compares the same market over time. The firm estimates that global consumer AI spending jumped from $12 billion to $40 billion in one year.
Usage barely moved at the same speed. The share of U.S. adults using AI increased from 61% to 64%, while Menlo estimates the global user base grew from roughly 1.8 billion to 2.0 billion people.
Spending more than tripled while the estimated number of users rose only 11%.
That tells us something more interesting than another adoption headline: monetization is finally catching up with reach.
Menlo found that 55% of AI users now pay for at least one AI product. Among those payers, 46% spend more than they did a year ago and only 12% spend less. The top 14% of payers generate 60% of consumer AI expenditure, so the market still depends heavily on a relatively small group of intensive users.
The $40 billion figure should be treated as an estimate. Menlo surveyed just over 5,000 U.S. adults and then adjusted the results to estimate global spending, accounting for differences in internet access and regional purchasing power.
Even with that caveat, consumer AI reaching this scale in only a few years is remarkable. Menlo calculates that app stores took about eight years to reach a similar level of consumer spending, while U.S. streaming took more than a decade.
How much are companies spending on generative AI?
Companies spent roughly $37 billion on generative AI in 2025 according to Menlo Ventures, more than three times the previous year and roughly twenty times the level measured in 2023.
Menlo estimates enterprise GenAI spending rose from about $1.7 billion in 2023 to $11.5 billion in 2024 and then $37 billion in 2025.
Few enterprise software categories have gone from almost nothing to that scale so quickly.
The methodology also makes this number more useful than a top-down forecast. Menlo surveyed around 500 U.S. enterprise decision-makers and combined their answers with a bottom-up model covering foundation-model APIs, model training, AI infrastructure software and applications.
There is another change underneath the headline growth. Menlo found that 76% of enterprise AI solutions are now purchased from vendors rather than built internally.
That is a big departure from the first GenAI wave, when companies were running pilots, wiring APIs together themselves and asking internal engineering teams to experiment with models.
These days, more companies appear willing to pay for finished products.
Microsoft’s latest numbers point in the same direction from a completely different dataset. Microsoft 365 Copilot has passed 30 million paid seats. Net paid-seat additions more than doubled sequentially in its latest quarter. The company’s broader first-party Copilot family now has more than 150 million monthly active users.
Enterprise GenAI is still small beside the full software industry, but calling it experimental no longer matches what companies are actually buying.
Is this market still open or already locked up? Our generative AI market report has the answer.
Where is all that enterprise GenAI money going?
Enterprise generative AI spending has moved up the stack surprisingly quickly, with applications already taking more money than foundation models, training and AI infrastructure software combined in Menlo’s 2025 estimate.
Menlo calculates that enterprises spent about $19 billion on AI applications during 2025. Foundation models captured $12.5 billion, model training $4 billion and AI infrastructure software another $1.5 billion.
That distribution is worth watching because the first years of the boom created a widespread assumption that model providers and compute companies would absorb most of the economic value.
Applications are already challenging that assumption.
Horizontal AI products took roughly $8.4 billion. Department-specific applications such as coding, sales and customer support generated another $7.3 billion. Vertical applications built specifically for industries such as healthcare and legal reached $3.5 billion.
The application layer went from roughly $600 million of enterprise spending in 2023 to about $19 billion two years later.
| Enterprise GenAI category | 2023 | 2024 | 2025 |
|---|---|---|---|
| Foundation models | $0.5B | $3.5B | $12.5B |
| Model training | $0.5B | $3.0B | $4.0B |
| AI infrastructure software | $0.1B | $0.4B | $1.5B |
| Horizontal AI applications | $0.3B | $1.6B | $8.4B |
| Departmental AI applications | $0.2B | $1.8B | $7.3B |
| Vertical AI applications | $0.1B | $1.2B | $3.5B |
| Total | ~$1.7B | ~$11.5B | ~$37B |
If you want more recent data on this point, please see our latest generative AI market report.
Is AI coding already a big market on its own?
AI coding has become generative AI’s clearest large standalone application market, with billions of dollars of direct enterprise spending and tens of millions of users.
Menlo estimates that enterprises spent around $4 billion on AI coding products in 2025. Coding alone represented roughly 55% of all departmental AI application spending in its study.
The company-level numbers have since become even harder to dismiss.
GitHub Copilot now has about 50 million users according to Microsoft. Anthropic’s extraordinary revenue acceleration has been driven partly by Claude Code and broader developer adoption. OpenAI’s recent revenue growth has also been helped by coding products.
Coding works particularly well for generative AI because the product can be measured against real output. Developers can ask whether a tool writes usable code, fixes bugs, completes migrations or saves hours of work. That is much easier to price than a vague promise that an AI assistant will make an entire organization more productive.
Usage can also become enormous. A coding agent may read thousands of files, repeatedly call models, run tests, inspect errors and try again before finishing one task.
This is one reason cheaper inference has not automatically translated into lower AI revenue. Coding agents can consume far more model calls as they become capable of doing more useful work.
Private startups are closing in on listed giants faster than most people think. See who's next in our generative AI market report.
Are OpenAI, Anthropic and Google taking almost all the model revenue?
The enterprise model market remains heavily concentrated today, even while users increasingly switch between several AI products.
Menlo estimated that Anthropic, OpenAI and Google together captured about 88% of enterprise LLM API spending in 2025. Its model put Anthropic at 40%, OpenAI at 27% and Google at 21%.
The exact percentages will keep moving quickly, but the concentration itself is hard to miss.
At the same time, customer behavior looks much less exclusive than those revenue shares suggest.
Menlo’s latest consumer study found that the average AI user now uses around three general-purpose assistants, up from 2.2 a year earlier. ChatGPT reached 60% of AI users in the survey, Gemini 58% and Claude 20%. The percentages overlap because the same person may regularly use several products.
Enterprise buyers behave similarly. Microsoft says more than 10,000 customers have already used more than one model through Microsoft Foundry.
Google, meanwhile, says more than nine million developers now build with its models every month. Gemini APIs recently reached approximately 22 billion tokens per minute, up from 16 billion just one quarter earlier.
So the model market currently combines concentrated revenue with surprisingly fluid usage. Customers have clear favorites, but many are still willing to move workloads around when another model becomes cheaper, faster or better at a particular task.
Are generative AI market estimates counting the same money several times?
Yes. Double counting can make the generative AI economy look much bigger than the amount of money actually coming from end customers.
Imagine that a company spends $100 with an AI coding startup. The startup records that $100 as revenue. It might then spend $30 with Anthropic for Claude inference. Anthropic records the $30 as revenue too. Anthropic may spend part of that money with Amazon or Google for compute, creating another company’s revenue. Those cloud providers buy chips and servers from suppliers further down the chain.
Adding every transaction would produce much more than $100 of vendor revenue even though the original customer spent only $100.
Real relationships make this problem especially important in AI.
Anthropic sells directly to companies but also reaches customers through AWS, Google Cloud and Microsoft. OpenAI sells subscriptions and APIs while buying enormous amounts of cloud capacity. Microsoft sells its own Copilots and also hosts third-party models. Amazon sells Bedrock access while providing infrastructure to several AI labs.
This overlap means we cannot calculate the true GenAI market by simply adding every AI-related revenue disclosure we can find.
It also explains why seemingly credible market estimates can differ by hundreds of billions of dollars without either side making an arithmetic mistake. One researcher may be measuring end-user spending while another is measuring revenue throughout the supply chain.
For the size of the commercial market, end-customer spending is the cleaner number. Supply-chain revenue tells us something different: how much economic activity the AI boom is generating across technology companies.
If you want more recent data on this point, please see our latest generative AI market report.
The gap between countries is bigger than most people think. See it in our generative AI market report.
Why is AI infrastructure already so much bigger than AI software revenue?
AI infrastructure spending currently dwarfs direct generative AI revenue because hyperscalers are building capacity for workloads they expect to arrive over several years.
IDC estimates that worldwide AI infrastructure spending reached $318 billion in 2025, more than double the $153 billion recorded one year earlier. Spending hit nearly $90 billion in the fourth quarter alone.
Gartner uses a broader definition and now forecasts about $1.48 trillion of AI infrastructure spending for 2026.
The company spending plans tell the same story. Amazon expects roughly $220 billion of capital expenditure this year. Alphabet recently raised its expected range to roughly $195–205 billion. Microsoft’s calendar-year figure is around $175 billion after adjusting for its accounting treatment of leases. Meta expects around $130–145 billion.
Those four companies alone are therefore heading toward roughly $720–745 billion of annual capital expenditure.
Every dollar in those budgets should not be labelled GenAI spending. The companies are building data centers, networks and cloud infrastructure that support many workloads. But AI is explicitly driving a large part of the increase.
The comparison with direct model revenue is striking. OpenAI and Anthropic’s latest reported run rates add to more than $105 billion. The four largest hyperscalers are spending roughly seven times that amount on annual capex.
The market is being built for a much larger future workload than the one customers pay for today.
| Current infrastructure measure | Approximate scale | What we learn from it |
|---|---|---|
| Amazon annual capex plan | ~$220B | AWS and AI capacity are major drivers |
| Alphabet annual capex plan | ~$195–205B | AI demand has pushed spending higher |
| Microsoft calendar-year estimate | ~$175B | Heavy cloud and AI capacity expansion |
| Meta annual capex plan | ~$130–145B | AI infrastructure dominates investment |
| Combined four-company capex | ~$720–745B | The buildout is enormous even beside AI revenue |
| IDC 2025 AI infrastructure spending | $318B | More than doubled in one year |
| Gartner 2026 AI infrastructure forecast | ~$1.48T | Infrastructure is the biggest part of its AI market |
Is Big Tech actually making enough AI revenue to justify all this spending?
Big Tech is already generating tens of billions of dollars from AI, but current revenue still leaves a huge amount of future growth baked into today’s infrastructure buildout.
Microsoft said earlier this year that its AI business had passed a $37 billion annual revenue run rate, growing 123% year over year.
Amazon has now added another clean data point. In its latest quarterly results, CEO Andy Jassy said AWS’s AI business had passed a $25 billion annual revenue run rate and was growing at triple-digit percentages. Amazon’s separate chips business also crossed $25 billion.
Google does not give us one equivalent AI revenue number, but its latest quarter offers several useful clues. Google Cloud revenue grew 82% year over year to $24.8 billion, with Alphabet explicitly pointing to AI infrastructure and AI solutions as major drivers. Nearly 90% of the Fortune 100 now use Gemini Enterprise.
The usage underneath that revenue is also climbing fast. Google says Gemini APIs moved from 16 billion to around 22 billion tokens per minute in a single quarter.
These figures make the AI infrastructure boom easier to understand. Customers are spending real money, AI workloads are growing quickly, and Amazon and Google both say parts of their infrastructure remain capacity constrained.
Still, hundreds of billions in annual capex require much more than today’s revenue. Data centers are long-lived assets, and the business case assumes that inference, agents, enterprise deployments and new AI products keep growing for years.
That assumption may prove correct. It is also where much of the financial risk sits.
Curious whether the hype matches reality? Our generative AI market report puts the trends next to the deals.
Could cheaper AI models make the generative AI market smaller?
Cheaper inference is currently expanding AI consumption faster than it is destroying revenue, because lower prices make many more tasks worth automating.
The fall in model costs has been extraordinary. Stanford’s AI Index found that the price of inference at roughly GPT-3.5-level performance fell from about $20 per million tokens in late 2022 to around seven cents by late 2024.
For a normal software product, a price collapse of that magnitude would look terrifying.
AI usage has responded by exploding.
Google’s model APIs are now processing roughly 22 billion tokens per minute. Agentic coding products may call models dozens or hundreds of times to complete one user request. Reasoning models can spend far more compute thinking through a problem than older chatbots used to generate a short response.
So one token becoming cheaper tells us very little about what happens to the total bill.
The economic unit is gradually changing too. Companies care less about the price of one million tokens when an AI system can complete a task that previously required an employee to spend several hours on it. The relevant comparison becomes the cost of completing the job.
Coding is already showing how this can work. Lower inference costs make longer-running agents economical, which increases total usage and creates products that could barely exist when every model call was expensive.
If model prices keep falling while the amount of useful work AI can perform keeps rising, the market can become much larger even as its basic computational unit gets dramatically cheaper.
If you want more recent data on this point, please see our latest generative AI market report.
How many people are actually using generative AI at scale?
Generative AI already has billion-user distribution, so the main consumer question today is how deeply those users engage and how many eventually pay.
Menlo estimates that around two billion people worldwide now use consumer AI products.
Individual platforms are already approaching that scale on their own. Google says the Gemini app has reached roughly 950 million monthly active users, with daily active users tripling over the past year. OpenAI has said ChatGPT reaches more than one billion weekly active users.
Microsoft offers another useful angle because many of its users are inside paid workplace products. Microsoft 365 Copilot has passed 30 million paid seats. GitHub Copilot has around 50 million users. Microsoft says its wider first-party Copilot family now exceeds 150 million monthly active users.
These numbers use different definitions, so lining them up as a ranking would be misleading. A weekly ChatGPT user, a paid Microsoft seat and a monthly Gemini user are different things.
Together, though, they settle the adoption question pretty decisively. Generative AI has already reached mass-market distribution.
What we still do not know is how valuable the average user eventually becomes.
Menlo’s consumer research illustrates the gap. Around 64% of U.S. adults in its survey use AI, while 55% of AI users pay for at least one AI product. Spending is heavily concentrated among power users. That leaves enormous room for revenue to grow if occasional users begin paying, but it also means the headline user counts currently overstate the depth of monetization.
Who's who in generative AI, at a glance. Our report goes deeper into each player and where they stand.
How much AI spending is genuinely new money?
Generative AI is creating large new revenue pools, but part of the apparent market growth comes from AI being bundled into software, cloud services and devices people were already buying.
The genuinely new categories are easy to see. Frontier-model APIs barely existed a few years ago. Neither did mass-market AI coding subscriptions, generative video platforms or most AI-native legal, healthcare and customer-support products.
Other spending is harder to separate.
Microsoft can add Copilot features to an enterprise software bundle and raise the price. Part of the increase reflects new AI value, but counting the entire Microsoft 365 contract as AI spending would obviously overstate the market.
The same issue appears in hardware. Gartner previously counted hundreds of billions of dollars of AI-capable devices inside its GenAI spending forecast. Yet a consumer buying a premium smartphone might have purchased roughly the same device even if the manufacturer had never advertised its on-device AI features.
Gartner’s newest forecast shows this blending becoming even more important. The firm says software companies are increasingly embedding agents and assistants directly into products customers already use. Companies are also choosing those ready-made features instead of building large custom AI projects.
AI revenue will consequently become harder to isolate over time.
Some of the biggest winners may stop reporting a clean standalone “AI revenue” number because AI will simply become part of search, cloud computing, productivity software, cybersecurity, CRM and other existing markets.
For anyone sizing GenAI, that means we should resist automatically treating every dollar of an AI-enabled product as incremental AI demand.
So how big is the generative AI market really?
The direct generative AI market is already in the low hundreds of billions of dollars today, while the broader economic ecosystem being built around AI is well into the trillions.
We can be more confident about the order of magnitude than about an exact figure.
OpenAI and Anthropic alone recently reached more than $105 billion of combined annualized revenue. Amazon says its AWS AI business has passed a $25 billion run rate. Microsoft previously reported more than $37 billion for its AI business. Menlo estimates consumers now spend around $40 billion on AI products, after estimating $37 billion of enterprise GenAI spending in 2025.
We cannot add those figures together mechanically. Some cover different periods, some measure annualized revenue, some overlap, and some include broader AI products.
But they put a clear floor under the market.
A direct GenAI industry worth only $40 billion or $50 billion no longer fits the observable revenue. At the other extreme, calling generative AI a $2.7 trillion end market today stretches the definition until servers, chips, consulting, conventional software and AI-enabled devices dominate the total.
The most defensible reading currently sits between those extremes. Customers are already spending well into the hundreds of billions across models and AI products, with the precise total obscured by private-company reporting, bundling and double counting.
Around that direct market sits something much larger. Gartner now expects about $1.48 trillion of AI infrastructure spending alone. Amazon, Alphabet, Microsoft and Meta are collectively planning more than $700 billion of annual capital expenditure. Hardware suppliers, cloud platforms and consultants add further layers of economic activity.
That gap is the most revealing number in the entire article.
Generative AI has already become a very large commercial market. The technology industry is building infrastructure for something several times larger.
Whether demand actually grows into that infrastructure is now the question that matters most.
If you want more recent data on this point, please see our latest generative AI market report.
OUR METHODOLOGY
This analysis asks how big the generative AI market really is by separating direct customer spending from the much larger infrastructure and services economy surrounding it. We looked at consumer spending, enterprise spending, model-provider revenue, application revenue, usage, inference economics and infrastructure investment as distinct dimensions rather than forcing them into one headline number.
We prioritized the freshest observable evidence available. Company earnings, investor disclosures and regulatory filings were used for public-company revenue, usage and capital expenditure. For private companies such as OpenAI and Anthropic, we used company disclosures where available and established financial reporting for revenue run rates that are not published in detailed financial statements.
Annualized revenue run rates are treated as current commercial scale, not as trailing twelve-month revenue. They are useful because the market is changing fast, but we do not add them mechanically to annual spending estimates or to one another when the underlying businesses can overlap.
We used Menlo Ventures’ consumer and enterprise studies as two separate spending views. The consumer estimate is based on a U.S. survey adjusted to estimate global spending, while the enterprise analysis combines survey evidence with a bottom-up model covering foundation-model APIs, model training, AI infrastructure software and applications.
We kept end-customer demand separate from supply-chain activity wherever possible. The same dollar can appear first as application revenue, then as model-provider revenue, then as cloud revenue and finally as hardware revenue. That is why infrastructure, services and semiconductor spending are used to describe the scale of the AI economy, not simply added to direct GenAI demand.
Infrastructure spending is treated as a forward-looking capacity signal rather than proof of current GenAI revenue. IDC’s AI infrastructure data, Gartner’s broader AI spending forecast and the capital expenditure plans of Amazon, Alphabet, Microsoft and Meta show how much capacity the industry expects to need, but those investments also support workloads beyond generative AI.
We used usage and cost data to test the economics behind the market rather than to inflate the market-size calculation. User counts, paid seats, API token volumes and Stanford’s inference-cost data help explain whether lower prices are shrinking revenue or enabling much larger quantities of AI work.
Key sources used in the analysis include Gartner’s latest worldwide AI spending forecast, Menlo Ventures’ 2026 consumer AI study, Menlo Ventures’ 2025 enterprise GenAI study, IDC on worldwide AI infrastructure spending, Stanford HAI’s AI Index, Microsoft’s FY2026 Q3 results, Microsoft’s FY2026 Q4 earnings call, Amazon’s Q2 2026 results, Alphabet’s Q2 2026 CEO remarks, Alphabet’s Q2 2026 SEC filing, Meta’s Q2 2026 SEC filing, OpenAI’s commercial-scale disclosure, Bloomberg’s reporting on OpenAI’s revenue run rate, and Bloomberg’s reporting on Anthropic’s annualized revenue.
We track how fast challengers are gaining on the incumbents. Our generative AI market report covers every player worth watching.
