NEW: SPACEXAI is the fastest growing software company according to Ramp spend data. I find the following: - SpaceXAI's advantage in coding + its new consumer offering could accelerate business adoption of its models, much like what powered Anthropic's rise. - Continued growth in cybersecurity SaaS. - Multiple winners in AI dictation (Wispr Flow and Granola both take a spot on our list).

NEW: AI spend stutters in the latest Ramp AI Index. Per employee spend rose in Sept for the top 1% of business spenders but remains below highs in July. The top 1% of firms spent $9,568 per employee per month, down 3.1% from its July high of $9,873. The median firm spent $16

We’re building the Post-AI stack for finance teams of the future, I’m hiring two people to help get us there. @a16z recently dropped a great piece (“You Need a New CFO”) describing what’s changing. Yesterday’s finance teams spent their focus operating in Excel/Sheets, and had to fight for engineering resources, or to carve out time for strategic work. Tomorrow’s finance leaders will build the future for themselves. AI makes everyone go faster, but only domain experts with taste know what good looks like. We’re hiring a finance person who is also a data person. Or vice versa. Someone who deeply understands the operating levers of a business, can build and sell their model to the CFO, then automate the roll-forward with SQL + python. Someone equivalently comfortable in a coding agent as they are leading a MBR. And once you solve for @tryramp, you’ll bring what you learned to the product and operations for every business that trusts Ramp with their finances. And we’re hiring another @arakharazian. Well, there is no one like Ara. But similarly, we’re looking for a person who can tell the data stories Ara does, with a credible perspective on: “Okay, but what does this mean for a finance team?” This person has a track record of writing and thinking in public that CFOs can’t ignore. They have a strong perspective on how finance teams should work w/ AI wherever they are today: what to build vs. buy, what post-AI financial work looks like, how to build the right teams. And they should be excited to back it up with our data. This can be an career-defining role that's as big as they want to make it. We are hiring one person for both of these roles, and we will wait for the right person. If either of these is you -- reach out! Show me what you’ve built. Today is day 2765 and we are just getting started.

AI Spend Breakdown: Top 1% Drives 80% of Enterprise AI Costs Catch the full episode here: AI's Reality Check: RenMac Off-Script with @arakharazian #AI# #Business#

Podcasts frustrate me because they are fluffy and high-level. This is the most in-depth discussion on the framework that @tryramp have used to hire, monitor, and scale their sales team. - Why you should overpay reps - Why you should cancel forecasting calls - Why bankers make the best reps - Why comp plans are BS and you should not have one to start @max11free, SVP of Sales is truly epic here and my notes from the discussion below. -- 1. Why You Should Overpay Your Sales Reps The early bet was to pay SDRs well above market rate and expect output well above market in return. Reps earn 2X a standard SDR but book 40, 50 or 60 meetings, four to six times the usual target, so the unit economics still work. 2. You Have to Treat Hiring a Sales Team Like Moneyball: How to Find Hidden Talent Go-to-market recruiting should follow Billy Beane, targeting top sellers at companies that grew revenue past nine figures despite a low NPS score. Those reps had to create their own demand, work complex cycles with weak third-party reviews and sell at a premium to cheaper rivals. The number one or two seller there can thrive in a system like Ramp. Single biggest advice on finding hidden gem talent @rabois? 3. Why Bankers Make Great Sales Reps Bankers are programmed to work, and a nine-to-five is foreign to them. Before the company had brand recognition, a cold call to a CFO from someone who had just left Goldman TMT IB carried weight. Early sales teams lack time to teach cold call and email basics, so bankers kept the team in pure execution mode. Find this also @PeetsChad 4. Why I Advise Not to Have a Comp Plan Founders are advised to pay 100% OTE for at least the first two quarters instead of setting a comp plan. A quota set too low can be blown out in six months, leaving the company owing a seller multiple six figures. A target picked arbitrarily can also prove unattainable and stall momentum. Haha, interested to hear your thoughts on this one @Carles_Reina 5. Never Hire Sales Reps One at a Time Hiring one seller or SDR at a time leaves no way to benchmark performance, and comparing them against the founder is not reasonable. Onboarding is laborious and expensive, so hiring in pairs is better, bringing competition and collaboration and keeping reps from being lonely. 6. Why the Most Annoying Sales Reps Tend to Be the Most Successful Sellers who constantly ask about account strategy, pipeline and product during onboarding show a direct correlation with being incredible. Quiet reps, or those asking questions an AI or Notion bot could answer, usually give a weak signal, though the inverse is not always true. 7. What Should Sales Talent Focus on When Deciding Which Company to Join? Sellers joining an early-stage company should not index on equity, cash and comp, but on the market, the talent density of the business and the founders. Taking a VP title and lots of equity at a mediocre business growing 20 to 30% a year is the wrong trade. Biggest advice to talent on choosing which company to join @breeves08

Things that would change my view here: increased *paid* adoption of AI by consumers, which would increase the market for AI beyond what it is currently: highly concentrated enterprise use in tech and finance

I think the competition between OpenAI and Anthropic is sufficient to explain the decline in AI prices, but that's not really the point. Whether it's OpenAI or Anthropic or open source models, less spend going to AI would, I think very reasonably, be net-negative for AI infra. Unless if the price declines are met with increases in actual spend that can then be reinvested in infra, but we don't see that today, so I cannot support that view with data. Spend seems to be slowing.

A lot of people sharing this, but the decline is driven by price cuts at the frontier, in turn driven by competition between OpenAI and Anthropic. Not open source models, which are <5% of business spend. And if frontier models see a hit to their margins, that will depress their ability to reinvest in infra.

"AI labs officially launch frontier cyberdefense products to help others counter threats from frontier ai." This is a fantastic prediction and is also why I have low concerns re: AI safety. We used to buy antivirus with every computer. Countries attacked nuclear facilities (and still do!) with computer viruses (stuxnet). Cybersecurity catches up. We still have breaches, but most people don't need to think about it. In the future, AI will similarly be deployed to combat and defend security threats from AI. All in the background.

NEW: Top SaaS Vendors on Ramp (Oct 2026) We clearly see AI spend pulled in two different directions. On the downside, new and cheaper models are pushing AI spend down. Jev captured an entire percentage point of market adoption within a month (unheard of growth). And routers continue to gain. On the upside, the increased adoption of expensive image and video generation models will push AI spend up, like Higgsfield and Runway.

NEW: Companies are switching their top model more than ever. The monthly share of firms switching providers reached a record 8% in September. This is why: 1/ ARR is a dumb measure for AI spend. There is no "RR." 2/ Prices and margins will continue to fall for AI companies.

Firms are switching AI providers at the highest rate Ramp has seen. The monthly share of businesses switching their top model provider reached nearly 8% in September, measured as a three-month moving average. We last saw comparable switching when Anthropic launched Opus 4 and Sonnet 4 in May 2025. Rapid model releases and OpenAI’s price cuts are giving firms reasons to switch. The more willing and able firms are to switch, the less pricing power AI labs have. Another sign that AI models may be becoming commodities. Read more at our data blog:

@BrianDeckerSF Very little business spend goes through open models. first of all, because they're cheap. second of all, because even the businesses that use open models STILL spend much, much more on openai and anthropic.

NEW: AI Spend FALLS in the latest Ramp AI Index. Price cuts at the frontier, in addition to cheaper and more efficient models at standard and lite levels, are pushing down the cost of using AI. Open source models remain <5% of business spend. This decline in spend is driven almost exclusively by competition between OpenAI + Anthropic.

Thank you @tylercowen for recommending your readers visit Armenia. It's fun and special and worth a trip I don't know who you talked to for (8) but Russia does not protect Armenia and many Armenians do mind its influence in energy / trade / security, to put it lightly