Who’s Actually Paying for AI? A Deep Dive into Startup Spending and the Top 100 Gen AI Apps
🚀 Hook: The AI Spend Party Is Just Getting Started
Imagine walking into a dazzling tech gala where the spotlight is on AI—but the real intrigue lies behind the scenes. Who’s actually getting paid? Which startups have cracked the code to get real dollars from real users? That’s what today’s spotlight sheds light on: where AI meets actual budgets, and where the hype starts to bend toward value.
The Big Reveal: Who’s Paying for AI (and Where)
Andreessen Horowitz (a16z), in collaboration with fintech Mercury, just dropped their first AI Spending Report—and it throws a bright light on the deep end of AI adoption. ([TechCrunch][1])
They focused on the top 50 AI-native application layer companies—essentially, those whose core business is built around AI. Using real transaction data (yes, dollars moving), they slice into which AI tools are actually putting in work in startup stacks. ([TechCrunch][1])
Here’s what jumped out:
- OpenAI leads the pack in startup spending; Anthropic is the strong No. 2. ([TechCrunch][1])
- Replit is shining brightly in the vibe-coding arena (ranked No. 3). ([TechCrunch][1])
- Lovable, despite much buzz in consumer traffic, lags in enterprise use (ranked No. 18 in spend). ([TechCrunch][1])
- Tools like Cursor (#6) and Emergent (#48) are also showing serious uptake. ([TechCrunch][1])
- In a twist, Cognition (with its specialized coding tools) sits at No. 34. ([TechCrunch][1])
The analysts interpret this as a signal that startups are still shopping widely—there’s no one “AI stack” yet. And while copilots and augmenting tools are dominant now, there’s anticipation of a shift toward end-to-end agents in time. ([TechCrunch][1])
They also note the blending of consumer and enterprise tools: things like Canva, Midjourney, etc., are creeping from personal use into the office. ([TechCrunch][1])
The “Top 100 Gen AI Consumer Apps” — 5th Edition: Figures at a Glance
From the a16z “Top 100 Gen AI Consumer Apps — 5th Edition” report, here are the key figures and trends you should know, summarized in table form:
| Metric / Feature | Figure / Value | Notes & Context |
|---|---|---|
| New names on web list in this edition | 11 | Driven by traffic increases vs previous edition ([Andreessen Horowitz][2]) |
| New names on mobile list | 14 | Especially as App Stores crack down on ChatGPT imitators ([Andreessen Horowitz][2]) |
| Total newcomers including Google splitouts | 15 | Four Google products now tracked independently ([Andreessen Horowitz][2]) |
| Google Gemini traffic (relative to ChatGPT) | ~12 % | On web ranking, Gemini sees ~12% of ChatGPT’s visits ([Andreessen Horowitz][2]) |
| Mobile ranking: Gemini vs ChatGPT | ~half | Gemini has “nearly half as many MAUs” as ChatGPT on mobile ([Andreessen Horowitz][2]) |
| Grok mobile growth (July 2025) | ~40 % spike | After release of Grok 4 + avatar features ([Andreessen Horowitz][2]) |
| DeepSeek drop from peak (web) | >40 % | DeepSeek has suffered a steep decline on web views ([Andreessen Horowitz][2]) |
| DeepSeek falloff (mobile) | ~22 % | From its peak in mobile usage ([Andreessen Horowitz][2]) |
| Chinese-developed apps in mobile top 50 | ~22 of 50 | Heavy concentration in photo/video AI apps ([Andreessen Horowitz][2]) |
| “All Star” apps — appeared in all five web rankings | 14 companies | Includes ChatGPT, Perplexity, Midjourney, Veed, etc. ([Andreessen Horowitz][2]) |
| Among those 14 All Stars: those with proprietary models | 5 | Others use APIs/open or aggregate models ([Andreessen Horowitz][2]) |
| Countries of origin for All Stars | 5 (U.S., UK, Australia, China, France) | Despite global use, origins are concentrated ([Andreessen Horowitz][2]) |
In short: the Gen AI consumer landscape is maturing. Some consolidation, but still plenty of new faces rising. Traffic and usage are stabilizing, but the model (the business models, the user behavior) is still far from settled.
💬 What It All Means (aka My Take)
The real story here isn’t just which apps are winning—it’s how deeply AI is embedding itself into startup DNA. That we now have a spending report (not just a usage ranking) suggests we’re exiting the era of fancy demos and entering the era of paying use cases.
- Diversity is still king: There’s no dominant universal AI tool yet. Startups are mixing & matching, trying different copilots, note-takers, agents.
- Consumer → enterprise bleed: Apps that delight consumers are being adopted in workplaces, sometimes organically.
- Agentic future moving slowly: Copilots (assistants) dominate now. But as trust, safety, models, and tooling improve, full agents (tools that do things end-to-end) may take over.
- Global play matters: Chinese AI tools are making big waves globally (especially in visual/video space), and many top apps now originate outside the U.S.
- Economic validation matters: Traffic and buzz are not the same as dollars on the line. A tool is only sustainable when someone is willing to pay for it.
If you’re building an AI app (or betting on one), you might ask: “Is my target a delightful consumer experience, an enterprise slot, or a hybrid path?” Because the winners will likely straddle fluid boundaries.
Glossary
- AI-native: A company whose core product is built around generative AI capabilities, not just peripheral AI features.
- Copilot: An AI tool that augments human workflows (e.g. aid in drafting, summarizing, assisting) rather than fully replacing.
- Agent / Agentic flow: An AI that can take actions end-to-end (planning, executing tasks) with minimal human intervention.
- Vibe coding: A trend where non-technical or semi-technical users “code by vibe”—i.e. generate or scaffold apps, logic, layout with AI assistance rather than hand-crafting lines of code.
- All Stars: In the a16z report, the set of Gen AI consumer apps that have appeared in all web-ranking editions so far.
- MAU (Monthly Active Users): A metric for how many unique users actively use a product in a month.
- TAM (Total Addressable Market): The total revenue opportunity if a product captures 100% of its target market.
Sources:
- TechCrunch: A new a16z report looks at which AI companies startups are actually paying for ([TechCrunch][1])
- a16z: The Top 100 Gen AI Consumer Apps – 5th Edition ([Andreessen Horowitz][2])
| [1]: https://techcrunch.com/2025/10/02/a-new-a16z-report-looks-at-which-ai-companies-startups-are-actually-paying-for/?utm_source=chatgpt.com “A new a16z report looks at which AI companies startups are actually paying for | TechCrunch” |
| [2]: https://a16z.com/100-gen-ai-apps-5/ “The Top 100 Gen AI Consumer Apps - 5th Edition | Andreessen Horowitz” |
