This blog post captures the highlights of a recent conversation between two tech-industry veterans as they reflect on the whirlwind of late 2025 and look ahead to the 2026 planning season.
The 2026 Planning “Oh sh*t” Moment
It feels like it was October just two weeks ago. We’re coming off the back of trips to Europe, the chaos of AWS re:Invent, and the Thanksgiving blur. Suddenly, the year is over, and we’re all headlong into 2026 planning. For anyone in sales or tech, that means one thing: the “oh shit” moment of looking at next year’s quota.
But before we dive into the grind of next year, we have to look at the data. ‘Tis the season for annual reports, and the numbers coming out of the AI sector are, quite frankly, mind-blowing.
The $37 Billion Hockey Stick
We’ve been tracking three major reports: Open Router’s 1 Trillion Token Report, OpenAI’s State of Enterprise AI, and the Menlo Report. While Open Router is great for seeing what the “tinkerers” are doing, the Menlo Report gives us the real enterprise pulse.
The Menlo Report pegged the current state of the AI market at $37 billion. That represents roughly 6% of total software spend across the Total Addressable Market (TAM).
To put that in perspective:
- 2022: $0
- 2023: $1.7 billion
- 2024: $11.5 billion (6.8x increase)
- 2025: $37 billion (3.2x increase)
We now have 10 AI products generating over $1 billion in ARR and another 50 companies doing over $100 million—all in less than three years. In the old SaaS world, hitting $100 million ARR was the ultimate benchmark of success (think of how fast Wiz did it). Now, it’s becoming the baseline for the top 50 players in AI.
Startups vs. Incumbents: Who is Winning?
One of the most fascinating takeaways is where the money is going. In the application layer, startups are actually beating the incumbents.
“For every $3 spent on AI applications in a category, the startup is capturing $2 of that spend.”
The breakdown of the market looks like this:
- Foundational Models/Infrastructure: The massive heavy hitters.
- Horizontal AI: (e.g., ChatGPT, Claude) Still beating out specific agents.
- Departmental AI: (e.g., Sales, Finance, HR tools).
- Vertical AI: Highly specialized industry tools.
Startups are winning in areas like AI coding assistants (think Cursor vs. GitHub Copilot) and market research. However, incumbents still hold the “moat” when it comes to data science and data silos, simply because they already own the integrations and the underlying data.
The Retention Reality Check
While the growth is insane, the churn is real. Product-Led Growth (PLG) is working better for AI than it ever did for traditional SaaS, but it’s a double-edged sword. It’s incredibly easy to buy, but it’s just as easy to cancel.
- B2B SaaS: Typical Net Revenue Retention (NRR) is around 70%.
- AI Native (Under $50/mo): NRR drops to a staggering 23%.
- AI Native (Over $250/mo): NRR climbs back to 70%.
The “AI Wrapper” is dying. Once OpenAI or Anthropic decides your core product is now a “feature” of their model, it’s over. If your tool doesn’t improve the actual workflow or integrate deeply into the tech stack, users eventually realize they can just copy-paste from ChatGPT and save the subscription fee.
The New Sales Tech Stack: From Clay to Quick Suite
We’re seeing a massive shift in how we actually do our jobs. Productivity isn’t just about a better CRM anymore; it’s about Knowledge Worker Productivity.
1. Amazon Quick Suite
Amazon is rolling out “Quick Suite” (a mashup of Q Business and QuickSight). It’s an enterprise-grade productivity app that authenticates into your Slack, OneDrive, and SharePoint. Imagine asking a bot: “Tell me everything I did with Customer X this week,” and it pulls the meeting recordings, the Asana tasks, and the email threads into a perfect summary.
2. Clay for Data Enrichment
In the “old days,” we pulled Apollo or ZoomInfo lists and manually hunted for emails. Now, tools like Clay aggregate multiple sources, enrich the data, and build automated workflow pipelines that pipe directly into Salesforce.
3. RevOps engineering (The Vercel Model)
Look at what Vercel did: they used AI agents to handle internal triage in Salesforce. They went from 10 SDRs to 1 SDR, reallocating the other 9 to outbound roles. They’re using AI to analyze meeting transcripts to find out why they lost deals—sometimes revealing that a “technical win” was actually a “human loss” because the economic buyer wasn’t actually bought in.
The Human Element
Despite all the AI “slop” and automated Bacho-style LinkedIn messages, one thing remains true as we head into 2026: Nothing replaces face-to-face. The AI can summarize the meeting, write the account plan, and draft the email, but it can’t sit across the table from a Chief Customer Officer and build the trust required for a three-year partnership. Use the tools to clear the “cog” work, but keep the agency where it belongs—with the human.
Would you like me to refine this blog post for a specific platform, such as LinkedIn or a technical newsletter?