- AI Fire
- Posts
- 😬 Is the Enterprise AI Boom Peaking?
In partnership with
This might be nothing, or the first warning sign. Ramp says top AI-heavy companies cut spend per employee nearly 10% in August. Cheaper models are changing.
What's on FIRE 🔥
IN PARTNERSHIP WITH VULTR
Tables, charts, and diagrams can get lost when PDFs are reduced to plain text for RAG. Vultr's new open-weight models preserve that visual context by indexing document pages as images.
This is reflected in the ViDoRe V3 benchmark, with Prime at #2, Core at #4, and Flash leading the sub-1B category.
AI INSIGHTS
Ramp’s latest data suggests business AI adoption slowed in August, while the heaviest AI users actually cut spending, summer doldrums or a warning sign? See some main findings here first:
56% of Ramp customers paid for AI products, up only 0.4% month over month.
Among the top 1% of AI-spending companies, spend per employee fell 9.7%, from $7,976 to $7,205.
Average token prices dropped to about $0.68 per million tokens, down from a 2026 peak of $1.15.
Businesses are increasingly choosing cheaper models instead of always paying for the newest frontier releases.
OpenAI reached 39.8% business adoption, while Anthropic climbed to 43.8%.
Only 6.4% of AI-spending businesses used model-serving or inference platforms in August.
Ramp says August seasonality may explain part of the slowdown, and its dataset skews toward tech-heavy companies, so this is better treated as an early signal than proof that enterprise AI demand has peaked.
→ AI is getting cheaper faster than business usage is expanding, which is great for customers but potentially uncomfortable for model labs counting on rapidly rising enterprise spend.
PRESENTED BY HUBSPOT
The Future of AI in Marketing. Your Shortcut to Smarter, Faster Marketing.
Unlock a focused set of AI strategies built to streamline your work and maximize impact. This guide delivers the practical tactics and tools marketers need to start seeing results right away:
7 high-impact AI strategies to accelerate your marketing performance
Practical use cases for content creation, lead gen, and personalization
Expert insights into how top marketers are using AI today
A framework to evaluate and implement AI tools efficiently
Stay ahead of the curve with these top strategies AI helped develop for marketers, built for real-world results.
RECAP: HOW TO BUILD AI AD SYSTEM
Live section during our webinar
A big thank you to everyone who joined us live, tested ideas with us, dropped questions in the chat, and shared the ad problems you’re dealing with right now.
We showed you how to create an AI-powered ad workflow that learns from what you’ve already tested, so your next creative starts with more context, better angles, and fewer random guesses.
→ That creates a much more useful loop: research → create → test → learn → improve!
If you missed the session, you can still follow the full workflow from the beginning and build the system alongside us here:
🌟 We also pulled together the resources from the session so you don’t have to rebuild everything yourself. Inside the resource pack, you’ll find the Claude prompts and frameworks & HeyGen workflow and examples we used.
|
TODAY IN AI
AI HIGHLIGHTS
🍎 Apple finally went foldable. The new iPhone Duo stole the show, but the event also packed in iPhone 18 Pro, AirPods, Apple Watch updates. Catch up everything here.
💸 Straitly just launched publicly. It pays you cashback for using AI models, offering 5% on closed-source & 10% on open-source models across 177 models. Try it here.
⏳ GPT-6 Astra (Medium) is temporarily available in Arena Direct Mode for just 24 hours. If you want a clean hands-on test, try it before Sept 10 at 8am PT (highly rcm)
🤯 This idea is way better than a morning app. A creator prints her own GrokBot-powered newspaper daily. If doomscrolling owns your mornings, copy this instead.
🧰 You might be losing a ridiculous amount of time to small repetitive tasks. OpenAI’s new 16-plugin bundle is built around exactly that. Browse and steal a few workflows.
👀 An Anthropic researcher just quit over fears that self-improving AI could spiral beyond human control. His warning: labs are racing faster than safety can keep up.
😬 OpenAI put an AI “doomer” on its board. Paul Christiano openly says AI could become catastrophically uncontrollable, and now he has a seat on OpenAI’s board.
💰 Big AI Deal: Qualcomm landed a major AWS AI deal, giving Amazon warrants to buy 25M shares worth about $4B. Amazon could also purchase up to $60B in Qualcomm server chips.
NEW EMPOWERED AI TOOLS
📄 Free Webinar: Build your own AI agent from technical documents in minutes. Turn 200-page specs, scanned tables, handwritten notes, and drawings into production AI Agents. Join the webinar.
🖼️ ChatGPT Images 2.5 creates sharper, more realistic visuals faster, with better editing, richer textures, and much stronger control.
🛡️ Harden AIF protects AI coding agents by checking tool calls before they run, keeping your code and tool output safely on your machine.
💻 Mastra Factory lets AI agents take software issues from planning to implementation & pull-request review, all inside one workspace.
AI BREAKTHROUGH
Google DeepMind launched AlphaGenome Atlas, a free research platform that precomputes the predicted effect of all 9 billion possible single-letter DNA mutations in the human genome. Main details:
The Atlas is a massive 1-petabyte dataset, more than 30× larger than the AlphaFold Database.
It covers both the 2% of DNA that codes for proteins and the much harder-to-interpret 98% non-coding region that regulates gene activity.
Each mutation gets an AlphaGenome Variant Impact (AVI) score, helping researchers rank which DNA changes are most likely to matter.
Researchers can also see why a mutation scores highly, including predicted effects on gene expression, RNA splicing, chromatin accessibility, and proteins.
The database includes more than 2,500 recurring DNA sequence motifs, helping scientists understand which regulatory “patterns” a mutation may disrupt.
Researchers can search it through a website without coding, use the AlphaGenome API, or connect it to Google’s Antigravity scientific AI system.
It’s already being tested on real disease research. Broad Institute researchers used the Atlas to prioritize mutations in previously unsolved rare-disease cases, including identifying a potentially important mutation in the DNM1 gene.
We read your emails, comments, and poll replies daily
How would you rate today’s newsletter?Your feedback helps us create the best newsletter possible |
Hit reply and say Hello – we'd love to hear from you!
Like what you're reading? Forward it to friends, and they can sign up here.
Cheers,
The AI Fire Team
Reply