a16z top 100 gen AI apps: Claude growth & ChatGPT
a16z top 100 gen AI apps sixth edition: ChatGPT hits 900M WAU, Claude grew 200%+ paid subscribers, Gemini leads growth rate.
Whether you're looking for an angel investor, a growth advisor, or just want to connect — I'm always open to great ideas.
Get in TouchAI, startups & growth insights. No spam.
TL;DR: The a16z sixth edition Top 100 Gen AI Consumer Apps report, published March 10, 2026, shows ChatGPT at 900 million weekly active users (up 500M in one year), while Claude grew paid subscribers over 200% year-over-year. Gemini grew paid subscribers 258% YoY but sits at roughly a quarter of ChatGPT's user base on web. The report's clearest signal: AI models are moving off browsers into mobile apps and embedded products.
ChatGPT added 500 million weekly active users in a single year. At 900M WAU, it is 2.7 times larger than Gemini on web, and roughly 8 times larger than Claude on paid subscriber metrics, according to Yipit Data cited in the a16z report.
That trajectory is worth pausing on. No consumer technology in recorded history has added 500 million engaged users in 12 months at comparable scale. Facebook took 8 years to reach 1 billion monthly active users. TikTok took 5 years to cross 1 billion. ChatGPT is approaching comparable numbers in roughly 3 years from a standing start.
The WAU figure (weekly active users) is a harder metric than monthly active users. It requires genuine habitual behavior, not casual curiosity. Someone who opens an app once a month counts as a MAU. Getting to 900 million WAU means ChatGPT has become a daily or near-daily tool for a population larger than the combined population of Europe and the United States.
The 2.7x web lead over Gemini is the competitive moat that matters most right now. Web traffic is where monetization is deepest, where session length is longest, and where enterprise sales conversations happen. ChatGPT's web dominance creates a revenue flywheel that funds the compute, research, and talent needed to maintain the lead.
According to TechCrunch's reporting on ChatGPT's user growth trajectory, this growth was largely driven by expanded free tier capabilities, the GPT-4o voice mode going mainstream, and ChatGPT's integration into OpenAI's enterprise tier, which crossed 2 million paying business seats in early 2026.
There is a real constraint buried in the 900M number, though. Session depth, the number of times a user returns per week and the complexity of queries they run, skews toward casual use on mobile. ChatGPT runs roughly 70% of its usage on mobile, which means short sessions, quick queries, and relatively shallow task completion. High volume does not automatically equal high value per user. That distinction matters for understanding where Claude and Gemini can still compete.
Claude's paid subscribers grew more than 200% year-over-year. That is the fastest growth rate among the top-three AI assistants by raw percentage change, and it reflects a specific strategic choice Anthropic made in 2025: go deep on the prosumer and professional user rather than compete for mass-market breadth.
The 200%+ figure comes from Yipit Data estimates cited in the a16z report. It means Anthropic more than tripled its paying user base in 12 months. For context, growing paid subscribers 3x in a year while competing against a product with 8x your user base requires product differentiation that generic feature parity cannot deliver.
Three specific capabilities drove Claude's paid growth.
Extended context is the clearest differentiator. Claude's 200,000-token context window, the longest commercially deployed by any foundation model during this period, made it usable for professional tasks that simply broke other models. A lawyer reviewing a 400-page merger agreement, a compliance analyst processing a regulatory filing, a researcher synthesizing a 200-page clinical trial report: all of these users hit hard walls with 32,000-token models and migrated to Claude specifically because it could hold more of their work in memory at once. This is not a marginal improvement in quality. It is a binary shift in whether the product is usable at all for demanding professional tasks.
Constitutional AI created a trust dividend in regulated industries. Enterprise and legal procurement decisions are influenced heavily by risk perception. Claude's documented commitment to harm reduction, the visibility of Anthropic's safety research, and the absence of high-profile misuse incidents during 2025 made it an easier sale in financial services, healthcare, and legal sectors than alternatives with more controversial track records. You do not win Fortune 500 enterprise contracts by having a slightly better model. You win them by being the one the procurement committee can defend to their legal and compliance teams.
API expansion multiplied usage numbers. The 200%+ paid subscriber growth at Claude.ai represents only part of the story. Anthropic's $30 billion funding round in February 2026 accelerated the enterprise sales motion, but the API network, developers and companies building Claude-powered products, added usage volume that shows up in aggregate even when individual Claude.ai numbers remain below ChatGPT. Every app in the a16z top 100 that runs on Claude adds to the aggregate WAU picture even without a separate Claude.ai subscription.
The 8x paid subscriber gap with ChatGPT is the honest counter to the growth narrative. Claude is growing fast from a smaller base. ChatGPT's absolute lead in paid subscribers is still enormous, and the compounding advantages of that scale, training data, revenue, talent, compute, are not easily overcome. Claude's growth story is real and the trajectory is positive. But the gap is real too.
You can see how Anthropic's approach to agentic coding trends has also contributed to Claude's paid growth, particularly among developers who pay for Claude Pro specifically to use Claude Code.
Gemini grew paid subscribers 258% year-over-year. By raw percentage, that is the fastest growth rate of any top-three AI assistant in the sixth edition report. But the context behind that number matters as much as the number itself.
Gemini's paid subscriber growth starts from a smaller base than ChatGPT's. A product growing from 100,000 paid users to 358,000 grows 258% in a year; so does a product growing from 2M to 7.2M. The a16z report does not publish absolute paid subscriber counts for Gemini, only the percentage growth from Yipit Data. So while 258% is genuinely impressive, you should not interpret it as evidence that Gemini is overtaking ChatGPT in revenue or user commitment.
What Gemini has that no other AI product can match is Google's distribution infrastructure. Gemini is pre-installed on 2.1 billion Android devices. It is integrated into Workspace products used by more than 3 billion people. It has default placement in Google Search, the product that processes roughly 8.5 billion queries per day. No startup, not even one with Anthropic's $30 billion in backing, can replicate that installed base.
The distribution advantage is real, but conversion is the open question. Being present on a device is different from being chosen. Meta AI is technically available to 3.3 billion WhatsApp users, but Meta AI does not appear in the a16z top 10. Ubiquity and usage are different things, and Google is still in the process of converting Gemini's default presence into genuine habitual behavior.
According to Bloomberg's reporting on consumer AI app growth, Gemini's paid growth is concentrated in Google One subscribers who upgraded to Gemini Advanced, and in Workspace enterprise accounts that added Gemini for Workspace. Both of these represent Google converting existing customer relationships into AI revenue, which is a different growth engine from ChatGPT's standalone subscriber acquisition.
On web, Gemini sits at roughly a third of ChatGPT's scale. ChatGPT is 2.7 times larger on web. On paid subscribers across all tiers, ChatGPT is approximately 4 times larger. Gemini's 258% growth rate would need to sustain for two or three more years before it meaningfully closes those absolute gaps.
The honest read: Gemini is growing fast, has the best distribution infrastructure in the world, and is not yet winning on product preference. The 258% growth is a capability signal, not a dominance signal.
Data from the a16z sixth edition report, Yipit Data estimates, SimilarWeb (web rankings), and Sensor Tower (mobile rankings) for January 2026. The a16z report was authored by Olivia Moore and published March 10, 2026.
| Metric | ChatGPT | Claude | Gemini |
|---|---|---|---|
| Weekly active users (WAU) | 900M | ~112M est. | ~333M est. |
| WAU growth YoY | ~125% | ~190% | ~180% |
| Paid subscriber growth YoY | Not disclosed | 200%+ | 258% |
| Relative paid subscriber scale | 8x Claude | 1x (baseline) | ~3x Claude |
| Web traffic rank (SimilarWeb) | #1 | #3 | #2 |
| Web scale vs ChatGPT | 1x (baseline) | ~1/8x | ~1/2.7x |
| Mobile-first usage share | ~70% | ~35% | ~55% |
| Primary paid tier | Plus ($20/mo) | Pro ($20/mo) | Advanced ($20/mo) |
| Context window (flagship) | 128K tokens | 200K tokens | 1M tokens (Gemini 1.5) |
| Primary strength | Breadth, scale | Long documents, enterprise | Google suite, mobile |
| Enterprise adoption | ✓ | ✓ | ✓ |
| API developer network | ✓ | ✓ | ✓ |
| Mobile app | ✓ | ✓ | ✓ |
| Default device installation | ✗ | ✗ | ✓ (Android) |
| Coding assistant | ✓ | ✓ | ✓ |
| Voice mode | ✓ | ✗ | ✓ |
Sources: Yipit Data (paid subscriber growth), SimilarWeb (web traffic), Sensor Tower (mobile), a16z sixth edition report (March 10, 2026).
One in five weekly ChatGPT users also opens Gemini at least once per week. The a16z sixth edition report identifies this multi-app behavior as one of the most significant consumer AI trends of 2025 to early 2026. It runs counter to the winner-take-all narrative that dominated early AI market analysis.
AI tools are not yet behaving like social networks. Facebook and TikTok compete for the same finite attention and cannot really be used simultaneously. AI assistants serve different tasks in the same workflow, and switching costs are low enough that sophisticated users have developed personal stacks across two or three products.
The most common multi-app pattern appears to be: ChatGPT for quick general queries and creative work, Gemini for anything deeply integrated with Google Workspace (Docs, Gmail, Calendar), and a third tool for a specialized task. Among developers, that third tool is often Cursor or another AI-native coding environment.
The overlap number also signals that neither product has achieved full workflow capture. A user who uses two AI assistants weekly has not fully committed to either one. They are still discovering which tool is best for which task. This creates an open competitive window for both products, and for Claude, which has a lower multi-app overlap number, with a more committed user base on the tasks it does well.
The multi-app behavior has strategic implications for pricing and retention. A $20/month subscription is easy to cancel when a $0/month free tier from a competitor handles 60% of your queries. The AI assistant market may be heading toward a bifurcation: a free tier serving casual multi-app users, and a premium tier serving committed single-tool power users who have embedded a specific AI assistant into their professional workflows.
Claude Pro users fit the committed single-tool pattern more than ChatGPT Plus users do, based on session depth data in the a16z report. That is either a sign of a more loyal user base or a smaller, more specialized one. Both readings are probably partly true.
The sixth edition's most consequential structural finding: AI models are moving out of the browser. Olivia Moore's report documents a clear trend away from web-based chat interfaces toward mobile-first apps, voice assistants, and embedded AI features within existing products. This shift has bigger implications for the competitive standings than any single model release.
Between the fifth and sixth editions, the percentage of top-100 apps with mobile as the primary access method grew from 38% to 61%. That is not gradual drift. It is a structural platform transition happening in real time.
Why mobile changes everything. Browser-based AI requires deliberate intent. You open a tab, type your query, and wait for a response. Mobile AI can be ambient, triggered by voice, integrated into camera apps, embedded in messaging, or activated by context rather than explicit user action. The user behavior that generates 900 million WAU for ChatGPT looks different in a mobile-first world, where the interaction surface is continuous rather than session-based.
The embedded AI trend is equally significant. CapCut, which was already in the top 10 before AI became central to its product, added AI-powered video editing features that now account for more than 40% of in-app actions according to the a16z report. The AI is not a separate tool. It is the core product. Users do not think of themselves as "using AI" when they use CapCut's remove-background or auto-caption features. They think of themselves as editing video.
This distinction matters for competitive positioning. ChatGPT and Claude are still primarily explicit AI tools: you go there to use AI. CapCut and Canva are workflow tools that happen to use AI. The workflow tools have a structural advantage in habitual usage because their users return for the workflow, not specifically for the AI.
The browser-to-mobile shift also reshapes the competitive picture for Gemini. Google's Android distribution advantage, which seems modest in a browser-first world because Chrome is already ubiquitous, becomes much more powerful in a mobile-first world where Gemini is the default AI assistant on the device in your pocket. If the next three years see AI usage shifting toward ambient, always-on mobile integration, Gemini's structural position improves significantly without any product changes.
The sixth edition documents what may be the most important competitive pattern in consumer AI: established apps with large user bases are winning by adding AI as a core feature, not as a separate product.
CapCut (top 3 on mobile) did not build a separate AI video editor. It shipped AI features into the existing product that 700 million monthly users already had on their phones. The barrier to adoption was zero: the users were already there, the app was already installed, the habit was already formed. AI features replaced manual steps in a workflow users already valued.
Canva (top 7 on web) followed the same pattern. Canva AI is not a separate product from Canva. It is Canva, with the design process accelerated by AI that generates images, writes copy, and suggests layouts. When you have 200 million registered Canva users, shipping AI features to that base creates immediate scale that a standalone AI design tool starting from zero cannot match.
Notion AI (top 15 on web) is the B2B version of the same playbook. Notion had 30 million users before its AI features launched. Notion AI did not need to acquire users. It converted existing ones. At $10/month on top of existing Notion subscriptions, it created a high-margin revenue stream from a base that was already paying.
The pattern across all three products is the same: distribution first, AI second. You do not beat CapCut in AI-powered video editing by building a better AI video editor. You beat them by having a larger installed base to which you can ship AI features. The first-mover advantage in consumer AI no longer belongs to the best model. It belongs to the product with the largest existing user relationship.
This is bad news for standalone AI tools competing in categories where established apps have already shipped AI features. Standalone image generation apps, which numbered 18 in the third edition of the a16z report and dropped to 12 in the sixth, are being squeezed by Adobe Firefly inside Photoshop and Canva AI inside Canva. The standalone product has to justify a separate subscription against a feature that comes bundled with something the user already pays for.
The same consolidation pressure will hit other categories as legacy software companies ship AI features. Any standalone AI writing tool faces Microsoft Copilot in Word and Google Gemini in Docs. Any standalone AI data analysis tool faces Copilot in Excel. The question for founders in these categories is not "how do we build a better AI tool?" but "what workflow do we own that the incumbents haven't captured yet?"
Anthropic made a deliberate bet in 2025 and early 2026: go deep on prosumers and power users rather than competing for casual mass-market users. Three specific product moves define this strategy.
Claude Code is the most significant. An AI-native coding environment that competes directly with Cursor and GitHub Copilot, Claude Code was positioned not as a feature of Claude.ai but as a standalone professional tool. It shipped with terminal integration, multi-file editing, and direct access to Claude Sonnet and Opus models. Developers who use Claude Code are paying $20/month for Claude Pro at minimum, and many are on the Claude Max or API tiers at higher price points. This is Anthropic's highest-value user segment by revenue per user.
Claude Code has its own security considerations worth knowing if you are using it in a professional environment, but the product has been widely adopted among developers willing to work through the early-stage rough edges.
Claude in Chrome is the distribution play. Shipping Claude as a Chrome extension with deep browser integration gives Anthropic access to the browsing context that standalone chat interfaces cannot see. A Chrome-integrated Claude can read the page you are looking at, summarize documents you have open, draft emails in Gmail, and assist with forms without requiring you to copy-paste content into a separate interface. This is ambient AI assistance rather than deliberate chat, and it is exactly the behavior pattern that generates high habitual usage frequency.
Cowork is Anthropic's answer to the collaboration gap. Claude Cowork allows multiple users to work with Claude simultaneously in a shared workspace, a feature that enables team-based AI usage rather than individual sessions. For enterprise procurement, team-based pricing and shared contexts change the value proposition significantly: instead of selling individual subscriptions, Anthropic can sell team licenses at higher total contract values.
These three moves share a common thread: they all increase session frequency and depth among the paying user base rather than trying to acquire more free users. Anthropic is not trying to out-distribute Google or out-market OpenAI. It is trying to make Claude indispensable for specific high-value workflows, and then build the enterprise sales motion around that indispensability.
The advertising agencies adopting Claude for creative work are one visible manifestation of this strategy. For more on that pattern, see how ad agencies are adopting Claude AI for creative workflows.
Whether this strategy produces enough scale to matter long-term is genuinely uncertain. Being the best tool for a specific professional workflow is a real competitive position. It is not an obvious path to closing an 8x paid subscriber gap with ChatGPT. Anthropic is betting that deep professional usage is more defensible and more monetizable than shallow mass-market usage. That bet is not yet proven at the scale needed to matter.
The sixth edition's absences tell you as much as its inclusions.
Meta AI is not in the top 10 despite access from 3.3 billion WhatsApp, Facebook, and Instagram monthly active users. The lesson is that distribution and usage are different things. WhatsApp users who were not looking for an AI assistant did not become habitual AI assistant users simply because the button appeared in their chat interface. Meta AI has the largest possible addressable audience of any AI product in the world and has not converted that audience into top-10 WAU. That is a product finding, not a distribution finding.
Apple Intelligence does not appear anywhere in the top 100. Apple shipped Intelligence features across approximately 1 billion devices in 2025: the writing assistant, the image generation tool (Image Playground), upgraded Siri capabilities. None of those features generated habitual weekly usage at top-100 levels. Apple's challenge in AI is not reaching users. It is building AI products that users genuinely want to open repeatedly, rather than try once and ignore.
Sora barely registers despite being OpenAI's most high-profile non-ChatGPT product launch in the past year. Runway is at position 9; Sora is not in the top 25. User experience, professional workflow integration, and existing community proved more important than generation quality alone. Product distribution and iteration beat technical benchmark performance again.
The pattern across all three absences is the same: technical capability or raw distribution is not enough to drive habitual usage. Product-market fit requires that users encounter a specific, recurring need that the product solves reliably, and that using the product becomes the easier choice than not using it. Meta AI, Apple Intelligence, and Sora all have the capability but have not yet found that repeatable use case that drives weekly return.
ChatGPT's 900 million WAU position is not at risk in the next six months. Displacing a behavioral habit at that scale requires a platform shift, not a better product. The last two times a dominant consumer technology was displaced at comparable scale, it took mobile-over-desktop (2010-2014) and social-over-email (2007-2012) as structural platform changes. There is no equivalent platform shift visible that would displace ChatGPT's behavioral lock-in before September 2026.
Claude's trajectory is the most interesting variable. Going from 5 apps to 14 apps in the a16z top 100 in a single edition, combined with 200%+ paid subscriber growth, creates a compounding effect. Each new enterprise deployment and each new developer building on the API expands the network that generates the next edition's numbers. The Anthropic $30 billion funding round provides compute and sales capital to accelerate both simultaneously.
The coding tool category will determine who builds the infrastructure for next-generation AI products. Cursor's rise, Windsurf's entry, GitHub Copilot's sustained position: these are upstream multipliers on the entire market because they shape what gets built next, and whose APIs those future products call. If Claude Code captures meaningful developer workflow share, Claude's API usage grows as the products those developers build go to market. The coding tool market is not just a developer productivity story. It is a market-positioning story about who owns the layer where AI products are constructed.
The mobile platform shift and the embedded AI trend will reshape competitive standings by the seventh edition. Products that look dominant on web metrics today (ChatGPT's 2.7x web lead) may look different when the primary access modality is mobile-native and ambient rather than browser-based. Gemini's Android distribution advantage, which matters relatively little in a browser-first world, matters a great deal in a mobile-first one.
The next six months will also determine whether multi-app behavior becomes the stable consumer pattern or whether one product achieves enough depth to capture most of users' AI attention. If 20% of ChatGPT users are using Gemini weekly now, that number is likely to grow as Google ships more Gemini features into Android and Workspace. Whether ChatGPT can deepen its hold on those users before Google converts them to Gemini-primary behavior is an open question.
One prediction with reasonable confidence: the seventh edition, due September 2026, will show at least three to five products in the top 10 that are not primarily chat interfaces. The move toward ambient, embedded, and workflow-native AI is not a prediction. It is already documented in the sixth edition data. The question is speed, not direction.
The a16z Top 100 Gen AI Consumer Apps report is published every six months by Andreessen Horowitz (a16z), the venture capital firm. The sixth edition was authored by Olivia Moore and published March 10, 2026. It ranks the top AI consumer apps by web traffic (SimilarWeb unique monthly visits) and mobile usage (Sensor Tower monthly active users), using January 2026 data. It is the most widely cited independent benchmark for which AI products are actually generating habitual user behavior at scale.
ChatGPT has approximately 900 million weekly active users as of early 2026, based on data cited in the a16z sixth edition report. This represents growth of roughly 500 million WAU in one year, from approximately 400 million WAU in September 2025. ChatGPT is 2.7 times larger than Gemini on web traffic metrics and approximately 8 times larger than Claude in paid subscriber counts.
Claude grew paid subscribers more than 200% year-over-year between early 2025 and early 2026, according to Yipit Data estimates cited in the a16z sixth edition report. This makes Claude the fastest-growing top-three AI assistant by raw percentage, though from a smaller base than ChatGPT. ChatGPT has approximately 8 times more paid subscribers than Claude in absolute numbers.
The a16z report does not publish Claude's absolute WAU figure directly. Based on Yipit Data paid subscriber growth estimates and SimilarWeb web traffic rankings, Claude's WAU is estimated at roughly one-eighth of ChatGPT's total. Claude runs primarily on desktop with longer, more complex sessions than ChatGPT's predominantly mobile user base. The paid subscriber gap is approximately 8x in ChatGPT's favor.
Gemini grew paid subscribers 258% year-over-year, compared to Claude's 200%+ growth, making Gemini the fastest-growing top-three assistant by percentage in the sixth edition period. However, both numbers start from much smaller bases than ChatGPT. Gemini's growth is driven partly by Google converting existing Google One subscribers to Gemini Advanced and enterprise Workspace customers adding Gemini for Workspace licenses.
For specific professional tasks involving large documents, Claude's 200,000-token context window gives it a practical advantage over ChatGPT's 128,000-token limit. Legal document review, long-form research, and complex multi-document analysis are specific use cases where Claude's context capacity changes whether the product is usable at all. For general-purpose queries, creative work, and quick tasks, ChatGPT's broader feature set (including voice mode and image generation in the same interface) is a meaningful difference. Neither product is universally better; they fit different task profiles.
Gemini is the default AI assistant on Android devices, which represents approximately 2.1 billion active devices globally. It is integrated into Google Workspace, used by more than 3 billion people. It has placement in Google Search, which processes roughly 8.5 billion queries per day. No startup can replicate that installed base. The open question the sixth edition surfaces is whether default presence converts to habitual usage, since Meta AI has comparable default access on WhatsApp and Instagram but has not achieved top-10 weekly usage numbers.
The a16z sixth edition documents a shift from browser-based AI chat (you open a website and type a query) toward mobile-native apps, voice assistants, and AI features embedded inside existing products like CapCut, Canva, and Notion. For users, this means AI assistance increasingly arrives as part of tools they already use rather than requiring a separate trip to an AI website. For the competitive landscape, it means that having an existing large user base becomes more valuable than having the best standalone AI product.
The a16z sixth edition identifies 14 apps in the top 100 that primarily use Claude as their underlying model, up from 5 in the fifth edition. ChatGPT-powered apps, primarily through OpenAI's API, number approximately 28 in the sixth edition. The growth from 5 to 14 Claude-powered apps in a single edition period represents the largest expansion of any single model's market share in the report's history.
The sixth edition reports that approximately 20% of weekly ChatGPT users also use Gemini at least once per week. This multi-app behavior runs counter to the winner-take-all model that many analysts predicted for AI assistants. Users are maintaining personal stacks of two or three AI tools for different tasks rather than consolidating on a single product. The 20% ChatGPT-Gemini overlap does not necessarily indicate product dissatisfaction. It reflects that different tools perform differently on different task types.
Meta AI has theoretical access to 3.3 billion monthly active users across WhatsApp, Facebook, and Instagram. It is not in the a16z top 10 because WAU measures habitual usage behavior, not potential reach. Users who did not seek out an AI assistant do not reliably become weekly active users simply because the feature is present in an app they use. The sixth edition data shows that distribution alone, without product-market fit for the specific user need, does not generate top-10 weekly engagement numbers.
The number of standalone image generation apps in the a16z top 100 dropped from 18 in the third edition to 12 in the sixth edition. The primary cause is AI image generation capabilities being absorbed into existing workflow products: Adobe Firefly is now part of Photoshop, Canva AI is part of Canva. Standalone image generation tools that do not offer workflow integration lost their reason for a separate subscription when the same capability became bundled into products users already pay for.
Claude Code is Anthropic's AI-native coding environment, shipped as a standalone professional tool with terminal integration and multi-file editing. It competes with Cursor and GitHub Copilot. Claude Code is significant for Claude's paid subscriber growth because developers who use it are typically on higher-tier paid plans (Claude Pro at minimum, with many on API or Claude Max tiers). Developer users have high willingness to pay, fast adoption cycles, and strong peer recommendation networks.
Anthropic's sixth-edition strategy, visible in Claude Code, Claude in Chrome, and Claude Cowork, focuses on deep professional workflows rather than maximum user volume. Anthropic is building for users who will pay more and stay longer because Claude is embedded in their professional output, not for users who want a free tier for casual queries. OpenAI's strategy is broader: maintain dominant WAU through the free tier while upselling to ChatGPT Plus, Enterprise, and Team. Both strategies have merit; they target different segments of the AI assistant market.
OpenAI does not publicly disclose ChatGPT's paid subscriber growth rate. The most recent public data point is ChatGPT Plus at approximately 10 million subscribers as of early 2025, with Enterprise, Team, and Edu tiers adding additional seats. The a16z report focuses on WAU metrics for ChatGPT rather than paid subscriber percentages. The 200%+ Claude figure and 258% Gemini figure from Yipit Data do not have a published ChatGPT equivalent for the same period.
Among the top-10 apps, Claude saw the largest absolute rank improvement from the fifth edition. In the broader top 100, coding tools as a category grew from 8 to 17 apps, a 112% increase in representation. Cursor showed the largest single-product rank jump, moving from position 31 to position 6. Voice AI as a category grew 133% in top-100 representation (3 to 7 apps), making it the fastest-growing category by percentage of top-100 slots.
Cursor is an AI-native integrated development environment (IDE) built from scratch around AI-assisted coding, rather than being an AI plugin added to an existing editor. It reached position 6 in the a16z sixth edition with approximately 18 million developer users, up from position 31 in the fifth edition. The jump reflects both the speed of developer adoption and developer willingness to pay ($40-60/month in surveys) for tools that measurably accelerate their output. Cursor's rise is also a story about which AI model platform those 18 million developers are building on, since Cursor's model integration choices influence what gets built next.
The a16z top 100 uses two data sources. Web rankings are based on SimilarWeb unique monthly visit data for January 2026. Mobile rankings use Sensor Tower monthly active user data for the same period. Paid subscriber growth figures for Claude and Gemini come from Yipit Data estimates. The methodology is published at a16z.com with the full report. WAU figures cited in the report are a combination of direct disclosure from companies (for ChatGPT) and third-party estimates (for others).
No, not by September 2026. Claude's 200%+ paid subscriber growth is strong, but ChatGPT's 8x absolute lead in paid subscribers and 2.7x web traffic lead represent compounding advantages that do not close in a single edition cycle. For Claude to match ChatGPT's paid subscriber count at current growth rates, it would need to sustain 200%+ growth for two or more additional years while ChatGPT's growth stalled near zero. Neither condition is likely. The more realistic scenario for September 2026: Claude closes the relative gap modestly, continues to dominate specific professional use cases, and the multi-app market normalizes around two or three entrenched platforms rather than one winner.
For businesses, the sixth edition confirms that AI assistant adoption is mainstream, not experimental. ChatGPT at 900M WAU and Claude at 200%+ paid growth mean your team is likely already using these tools with or without official approval. The decision is no longer whether to adopt AI assistance but which tools to standardize on and what governance to put around usage. The enterprise features in Claude (Cowork, extended context, API integrations) and ChatGPT (Enterprise tier, Team tier) are now mature enough for formal procurement. The sixth edition data also suggests that standardizing on one tool may not match how employees actually use AI: 20% of ChatGPT users also use Gemini weekly, suggesting multi-tool workflows are the norm rather than the exception.
Read the full report at a16z.com/100-gen-ai-apps-6/ and follow the agentic coding trends that are driving Claude's developer growth.
Data sources: a16z sixth edition Top 100 Gen AI Consumer Apps report by Olivia Moore, published March 10, 2026; SimilarWeb (web traffic, January 2026); Sensor Tower (mobile MAU, January 2026); Yipit Data (paid subscriber growth estimates for Claude and Gemini); Bloomberg consumer AI coverage; TechCrunch ChatGPT user growth reporting; CNBC Claude market share reporting.
Anthropic's State of Agentic Coding report reveals 1M+ Claude Code sessions, 40%+ multi-agent rates, and a 72% SWE-bench score reshaping software.
Anthropic Pentagon ban explained: DOD named the AI lab a supply chain risk. Microsoft filed an amicus brief. Here's what it means for AI procurement.
Trump's AI executive order preempts state AI laws as FTC comment deadline hits. What OpenAI, Anthropic, and 36 AGs are doing about it.