The End of B2B SaaS: AI Barbarians at the Gates and the Rise of the Mercenary Class

The Great B2B SaaS Era Is Winding Down

The End of B2B SaaS: AI Barbarians at the Gates and the Rise of the Mercenary Class

The Great B2B SaaS Era Is Winding Down

It’s not controversial to say that the last decade of American startups has been the era of B2B SaaS. The numbers back it up: according to Dealroom data, nearly half of all VC dollars have gone into SaaS, with B2B commanding a staggering 88% of that pie.

It’s been a great run — strong exits, intense competition, and an entire class of specialized investors built around it. But I’ve been increasingly convinced that AI is about to fundamentally reshape this landscape, and I want to lay out why.

Why VCs Fell in Love With B2B SaaS

Before we talk about what’s changing, let’s be honest about why B2B SaaS became the comfort food of venture capital. Sure, there’s technology innovation, but the real appeal? The structural advantages.

High margins, capital efficiency, scalability. Software doesn’t get meaningfully more expensive as you add customers. Cloud delivery eliminated geographic constraints. You get explosive growth on minimal upfront capital. That’s catnip for VCs.

Then there’s the lock-in effect. Once a B2B product is deeply integrated into a customer’s workflow, switching costs go through the roof. High retention means stable, compounding revenue on an existing customer base.

Predictability. Most B2B SaaS runs on Annual Recurring Revenue with 2+ year contracts. Compared to one-time sales models, this is infinitely more stable and forecastable. Lower churn, higher retention, strong upsell — it all gives investors a warm, fuzzy feeling about risk-adjusted returns.

Standardized due diligence. This might be the most underrated reason. B2B SaaS gave VCs a neat, standardized playbook: ARR, CAC, LTV, Retention Rate, Churn Rate, Net Dollar Retention. A handful of numbers that let you compare any two SaaS companies apples-to-apples. Even when deals fail, these metrics provide clear post-mortem justification. It reduces the pressure on individual VCs to have sharp instincts about trends — you just follow the formula.

Why B2B SaaS Bores Me

Here’s my contrarian take: the fact that these standardized metrics exist is precisely what killed the big opportunities in B2B SaaS.

The VCs who got rich were the ones who understood B2B SaaS dynamics before the playbook was public. Now? Every investor is staring at the same dashboards, running the same models. Founders themselves know their metrics cold and come to the negotiating table armed with comparable valuations.

Can you really “buy low, sell high” in a market where everyone has perfect information? Is evaluating pre-chewed SaaS metrics really helping VCs at the frontier of technology develop their edge?

I believe that in a perfect information game, the market converges on fair valuations. Outsized returns become structurally difficult. That’s why I think B2B SaaS with proven metrics is a relatively less attractive bet for early-stage VCs hunting for outliers.

AI Barbarians at the Gates

B2B SaaS has been the reliable gukbap (Korean comfort food — hearty, dependable, always there for you) of the venture world. But the AI invasion is an unavoidable event, and the convergence of AI and SaaS is the zeitgeist of this moment.

I’ve said this before and I’ll say it again: AI is a foundational technology that applies across every sector and every stage. It’s not a “sector” — you can’t put it in a box. B2B SaaS, meanwhile, is a business model, not a sector. These two aren’t competitors. They’re naturally complementary — a base technology meeting a business model.

Thinking of SaaS and AI as two separate sectors is like thinking of the internet and commerce as separate sectors, or mobile and transportation as separate sectors. The internet + commerce gave us e-commerce, online advertising, and social networks. Mobile + transportation gave us ride-sharing, DoorDash, and robotaxis.

I believe the AI + SaaS convergence will produce equally world-changing sectors. And it will rewrite the rules of the entire game.

That said — and I want to be clear — this doesn’t mean today’s AI companies are automatically the winners, or that current valuations are justified. The protagonists may or may not have appeared yet, and the valuation bubble is undeniable. I get the FOMO, believe me. But there’s a massive difference between mega-VCs and Big Tech who can spray money to improve their odds, and early-stage VCs like us who need to be surgical.

OK, enough preamble. Here’s what I think AI is about to do to B2B SaaS.

AI Will Break B2B SaaS’s Pricing Model, Margins, and Efficiency

The biggest structural difference between AI foundation models and traditional B2B SaaS? Pricing. Most AI models run on usage-based pricing.

Traditional SaaS thrived on seat-based fixed pricing built on cheap cloud costs. As AI gets baked in, that model stops working. Usage-based pricing means costs scale with customers. Margins compress. The old economics — high margins, cheap scaling — simply can’t hold.

For reference, B2C products like ChatGPT use flat subscription pricing but throttle usage when it gets too heavy. That’s fine for consumers, but it’s a non-starter for B2B where your enterprise client can’t have the product go down for hours — like ChatGPT literally did yesterday.

At the same time, usage-based pricing kills the long-term contract structure that made “recurring revenue” meaningful. Take OpenAI’s API revenue — calling it “recurring” is generous when there are no lock-in contracts and customers can migrate to another platform tomorrow. We’re already seeing startups hedge by working with multiple foundation models simultaneously (Perplexity, for example) or architecting for easy platform migration from day one.

Google is playing this brilliantly, by the way — making Gemini accessible through OpenAI’s own library so developers can switch with minimal friction. And honestly, if AI is as powerful as everyone claims, translating a system built for OpenAI to work with Microsoft, Google, or Amazon should be trivial. That’s the whole point, right?

So here’s the dilemma: AI is breaking the structural advantages that made B2B SaaS attractive — long-term contracts, high margins, cheap scalability. But SaaS companies that don’t adopt AI will get abandoned by customers. Damned if you do, damned if you don’t.

B2B Customers Will Accelerate the AI Cost Race to the Bottom

Here’s my spicy prediction: AI’s hallucination problem won’t be meaningfully solved within five years. That means full AI integration into high-accuracy B2B products is going to take longer than the hype suggests.

Evidence? Even Glean — one of Silicon Valley’s hottest enterprise search platforms — opened a workforce center in India to have humans re-verify AI outputs. It’s still cheaper than no AI at all, but it’s a far cry from the pure-AI margins and efficiency that justified their sky-high valuations.

The problem: B2B SaaS companies are under pressure to adopt AI now. Accuracy isn’t guaranteed. Costs are going up. The old formula is broken. (This is exactly why I think B2C AI will advance faster — consumers are more forgiving of errors.)

I expect this cost pressure, combined with margin compression and customers who still demand the fixed pricing they’re used to, will drive B2B companies toward open-source and smaller AI models. Companies will accept slightly lower performance in exchange for better privacy, faster speed, and lower costs that let them maintain fixed pricing for customers.

Competition will intensify. Costs will eventually crater. We might even circle back to fixed-cost models — but that transition could take longer than people think. Remember, the era of high license fee + low maintenance fee pricing lasted quite a while before the modern SaaS model emerged.

Bottom line: AI will disrupt B2B SaaS pricing in the short term, and the trajectory depends entirely on how fast AI costs come down.

The Rise of AI Mercenaries

Until now, B2B SaaS meant subscribing to specific tools and stacking them into your tech infrastructure. The coming era is different: companies will use AI to build the functionality they need, custom-fitted to their stack. (Honestly, this feels like Korea’s SI — systems integration — model making its way into the AI world.)

Case in point: Klarna, Sweden’s fintech giant, terminated its Salesforce and Workday contracts and is building internal replacements using AI.

Large enterprises might build internal teams for this. But for most companies, I think there’s about to be a market for something new: external organizations that don’t just handle one function, but use AI to customize and manage a company’s entire tech stack. High-productivity operators who stay independent — serving multiple clients, earning more, maintaining freedom — while delivering services at lower cost than traditional SaaS subscriptions.

Think of it as “AI operators” or “AI mercenary squads” — a modern, AI-powered version of the sales engineers who used to do massive Salesforce customizations for each client in the early days. Except now, AI makes it exponentially faster and more efficient. SaaS 2.0.

Their pricing would likely be task or function-based — what Silicon Valley is currently branding as “Service as a Software.” Given the variable cost structure of AI, I’d expect something like API cost + fixed fee.

Side note: I have two pet peeves about Silicon Valley trends. First, swapping words around and acting like it’s profound (Tech-bio, Service as a Software). Second, changing one letter in an existing word and calling it innovation. I get that rearranging words sounds hip, but “Service as a Software” doesn’t even convey its creator’s intended meaning of “task-based pricing” very well. Even in the Valley, it’s not landing with the impact they hoped for. Just my personal opinion — your mileage may vary.

Cut to the Chase

Here’s my bottom line: I believe B2B SaaS is heading into a period of significant disruption in pricing and competitive dynamics driven by AI. This isn’t the time for blind loyalty to a model that worked for the last decade. It’s the time to watch, question, and adapt.

And here’s where I’ll really go against the grain: I think the more attractive investment opportunity right now might actually be in B2C. Why? Consumers have higher tolerance for AI mistakes. B2C products can go viral and scale fast. Personalization — AI’s superpower — matters most in consumer products. And B2C is being overlooked right now because of recession fears. That’s exactly when you want to be paying attention.

Waiting for the B2C AI Killer App

Think about the greatest companies of our generation: Facebook, Google, Amazon, Netflix, Uber, Airbnb, Apple. What do they all have in common? They all started as B2C companies.

B2C markets have historically produced world-changing companies during major platform shifts — when both technology and distribution are simultaneously disrupted. The internet did it. Mobile did it. I believe AI is doing it again. This is the golden window for the next generation of B2C companies to emerge.

As an early-stage VC at the dawn of the AI era, I’m not looking for safe, steady B2B growers. I’m looking for the next outlier — the company we’ll all be using five years from now, the one that reshapes how we live. If you’re a founder with that kind of ambition, reach out to us at Sazze Partners.

Revenue? Returns? Market size? Technology? Pedigree? Sure, those matter. But what I’m really looking for in global founders is intelligence, vision, velocity, conviction, and influence. When everyone says the market is tough, when nobody else is swinging — that’s exactly when the opportunity is greatest.