AI's Impact on 2026 Budgets
Plus the latest benchmarks and insights from ICONIQ's State of AI report
Today’s Sponsor: Deel
Payroll is entering a new era with AI-driven compliance, real-time visibility, and higher expectations from the board.
Deel’s Payroll Strategy Toolkit gives finance leaders a step-by-step roadmap to modernize payroll, reduce manual risk, and build a scalable strategy with confidence.
AI Budgets Are Wide Open
Product/market fit means being in a good market with a product that can satisfy that market. — Marc Andreessen
Marc’s advice is spot on. The product is important, but there are many good products. You must also be in a good market.
Are you selling SaaS in an AI market?
CFO: You want budget for an AI product?
VP of Eng: Yes, it will increase productivity by 30%.
CFO: Incredible! Just tell me how much you need and I will make it work.
VP of Eng: Great! I also want to buy a SaaS tool. It’s only $500/month
CFO: No. There is no budget for that.
Every company has AI adoption goals, boards want to hear AI success stories, and CEOs are pushing leaders to show AI adoption in their departments. It is 100x easier to get AI budget than budget for another SaaS tool.
Guess where we are getting the AI budget from? Previous SaaS and headcount spend…
Current Customers Want to Buy From You
I don’t want to purchase from another vendor. I have too many apps already. I want to purchase AI stuff from my existing vendors. I bought from them because I trust them.
The AI opportunity is my current SaaS vendors’ opportunity to lose.
Prioritize selling your new AI stuff to your existing customers first. Not new logos. Most companies just send their CSMs or account managers to go upsell existing so sales reps can focus on new business. I think this is a mistake for many companies. AI features and their related pricing changes will feel like a whole new sales process. Don’t put your CSMs against a hot AI-native company’s sales reps for what feels like an entirely new sale. They will lose.
Re-win your existing customers first and then win new customers.
Finding Durable Moats
If technology was your moat, then you will likely be disappointed with your retention rates over the next year…
Everyone is searching for their durable moats that will withstand AI.
Nearly 70% of companies are building vertical AI applications, reinforcing that durable value is being created through domain-specific workflows rather than generalized intelligence.
There are riches in the niches. People are going deeper to differentiate.
The private equity folks are thinking the same. There has been a significant shift in PE activity over the past year between horizontal versus vertical software. PE is following where they think the most durable revenue will be.
No One Actually Wants to Build Internally
There continues to be lots of declarations that SaaS is dead because companies will just build all their apps internally with AI.
Listen. Almost no one wants to build these SaaS/AI applications internally. It almost always has terrible ROI.
Most of the “let’s just vibe code this” folks are quickly discovering that building a working version of a product is the easy part. It’s everything else that takes a lot of time and is risky.
The MoltBook Example:
Moltbook blew up on social media just a few days ago. Moltbook is a social network for AI agents only…Creepy, right? Within hours of launch, there were thousands (and then 1.5M+) AI agents on the platform. Talking about all sorts of crazy things — taking over the world, killing their human creators, inventing their own language so their humans couldn’t watch them, etc.
I tweeted the below somewhat jokingly, but was also very confident that some huge security concern would come from Moltbook.
Well…it didn’t take long for me to be right.
Wiz (cybersecurity company) found an unauthenticated database exposure issue that leaked 6,000 user emails, 1.5M API tokens, and private AI-agent messages. One reason for some of the crazy, viral “AI agent” posts was the security issue allowed humans to post as if they were AI agents…
As we see over and over again with vibe coding, although it runs very fast, many times people forget the basics of security. — Ami Luttwak, Wiz Cofounder
It only takes a few major security incidents from vibe coding projects to scare people away from trying to take on major projects internally. It’s rarely worth the risk. It’s almost always a low ROI activity. And it’s a distraction.
I would much rather have the ability to blame an external (widely trusted) vendor than have to explain why I tried to build my own ERP to save a few thousand dollars.
It’s not just security (although that will be the most visible). It’s also maintaining, supporting, updating, etc that can take so much time and distraction from what the team should be focused on. Maybe one day it will make sense for companies to build their own CRM, but 99.9% of the time it still doesn’t make sense today.
Everyone Going Multi-Product
AI is pushing everyone to go multi-product sooner and more broadly.
It’s much easier to build adjacent products now
Multi-product companies create stronger moats in the age of AI, which is reason enough to push multi-product.
A lot of your friendly partners in adjacent categories may become competitors much sooner than you expected. Prepare for that.
AI is Getting Budget…But Not Much Else
Building (and adopting) AI is getting budget in 2026. Everything else? Very very limited budget is available.
AI agents are eating SaaS TAM… Seats are disappearing and AI agents are achieving the desired outcomes.
Data & Process Readiness
A lot of companies are still not ready to fully leverage AI because their data and processes are messy. A lot of people complain that they aren’t seeing the value from AI agents. A big reason I see is because their data and/or processes (often both) suck.
Companies with bad data and processes need to have a 2026 (or even H1 2026) goal to fix their issues. Otherwise you will get left behind. Honestly this is most companies. The “mostly ready” bucket below are mostly just the optimistic companies.
Gross Margins
AI gross margins are increasing a lot…nearly 10 percentage points in all the different segments below.
These companies are already operating at the highest revenue per employee we have ever seen.
Revenue/FTE is up ~75% for the top decile of AI/software companies in just 2025🤯
So if gross margins continue to increase (and retention is actually strong), then these companies will be way better money printers than legacy SaaS because they won’t have the high stock-based comp dilution problem that SaaS companies have always had since they have a fraction of the employees.
Lots of Pricing Changes Are Coming
37% of companies plan on changing how they price AI products over the next year.
Pricing changes are really hard to get right. And I promise that a lot of customers and potential customers hate when you change things. You will lose customers if you get it wrong. Take your time and get it right…
Footnotes:
Great read: Payroll Strategy Toolkit. My friends (and today’s sponsor :) at Deel wrote the playbook for how payroll should work in 2026.
Want to Sponsor? I am opening up the rest of the year now so email me at onlycfo@onlycfo.io to have your wildest marketing dreams come true.
*nothing should be considered investment, legal, tax, or any other kind of advice.
Additional SaaS attack vectors that could be mentioned:
- Foundational models creating vertical integrations (e.g. Anthropic entering legal)
- Independent contractors (Fractional CAIOs) building out solutions for orgs
- Marginal cost of developing software will trend toward 0
Discussed some of these here: https://theinterneteconomy.xyz/p/ai-isnt-your-competitive-advantage