The death of SaaS is greatly exaggerated. The real story for emerging markets is the opposite: AI is making enterprise-grade software affordable and customizable for the first time.
I just started writing a few pieces a month ago, (on EdTech/AI, SaaS/AI & GTM), posting them only on my new site, with the posting date generic. https://alancyates.com/#writing I'll continue and shift where / how I publish...
Sharp and prophetic piece, Will. The line that hit hardest: "If I were a public equities buyer, I would buy historically low-priced SaaS companies."
One pattern worth adding to your thesis. The SaaS companies that are genuinely re-architecting their service around agents and customer outcomes are not just surviving the repricing. They are quietly outgrowing it.
HubSpot is the cleanest case. The stock is down roughly 70% from its 2025 high and the company is trading near multi-year lows. And yet Q4 2025 revenue grew 20% year-over-year to $847M, 2026 guidance came in above Wall Street expectations at $3.69–3.70B, and management announced a $1B buyback. What is hidden inside that growth is the shift in how the revenue is actually being earned. HubSpot's Breeze Customer Agent now resolves over 50% of support tickets autonomously. Agents account for roughly 60% of all credits consumed on the new consumption-based pricing layer. Over 8,000 customers have activated Customer Agent. Analysts estimate AI credits are already contributing a full percentage point to 2026 growth on a $3.7B base.
The per-seat story is softening. The agentic story is compounding. The market is pricing the first and missing the second.
This is why your buyer thesis holds. The companies re-articulating their service around what the agent accomplishes rather than how many humans log in are the ones that get to grow through the repricing. The ones still selling seats are the ones getting cut in half. Salesforce down 26% YTD with "seat compression" in the earnings commentary is the counterexample that proves your point.
Really interesting take, especially the Global South angle, thanks for sharing. That feels underexplored. I do think the “anti” case is a bit underweighted though.
IMO, It’s not that enterprises fully replace SaaS. It’s that SaaS gets hollowed out. Systems of record stay, but get pushed down to the “database layer” while workflows move into agentic layers around them. That compresses value per seat without killing the product. The “companies won’t build software” point also feels like it’s shifting. AI is lowering the cost enough that anything close to revenue or operational leverage starts getting built internally.
On valuations, strong earnings today don’t really disprove the bear case. Feels like the market is repricing future margins and pricing power, not just current revenue. On the Global South angle, I wonder if this is insulation or just lag. The same forces making software more accessible could also make it more competitive and commoditized locally. The key question for me is how the The Innovator's Dilemma plays out here with incumbents.
Hard agree on build + buy though. That feels like the real shape of things.Curious how you think about where value actually accrues in this stack.
I do not disagree with the price pressure point - it needs to and will happen. But I think it will take a long time for enterprises to start building mission critical workflow on top of price-pressured systems of record. Why? maintenance. Building s/w is easy compared to maintaining it in bug free, cyber-safe stable-state forever. For small companies or non-mission critical things, sure maybe that will happen more quickly. But repalcement of enterprise grade saas systems by the enterprises that have been locked into a complex feature set for years will not happen anytime quickly regardless of the state of vibe coding, IMNSHO.
Thanks, really insightful, especially on the maintenance point, particularly as someone building in the space. Completely agree that’s where things get real and why core systems stick; nuance being less replacement, more workflow shift, so pressure shows up in value per seat rather than displacement.
I just started writing a few pieces a month ago, (on EdTech/AI, SaaS/AI & GTM), posting them only on my new site, with the posting date generic. https://alancyates.com/#writing I'll continue and shift where / how I publish...
A similar view: https://alancyates.com/saas-evolution
Well said. We are on the same page. When did you post that? Sustack is not showing a date for some reason.
Sharp and prophetic piece, Will. The line that hit hardest: "If I were a public equities buyer, I would buy historically low-priced SaaS companies."
One pattern worth adding to your thesis. The SaaS companies that are genuinely re-architecting their service around agents and customer outcomes are not just surviving the repricing. They are quietly outgrowing it.
HubSpot is the cleanest case. The stock is down roughly 70% from its 2025 high and the company is trading near multi-year lows. And yet Q4 2025 revenue grew 20% year-over-year to $847M, 2026 guidance came in above Wall Street expectations at $3.69–3.70B, and management announced a $1B buyback. What is hidden inside that growth is the shift in how the revenue is actually being earned. HubSpot's Breeze Customer Agent now resolves over 50% of support tickets autonomously. Agents account for roughly 60% of all credits consumed on the new consumption-based pricing layer. Over 8,000 customers have activated Customer Agent. Analysts estimate AI credits are already contributing a full percentage point to 2026 growth on a $3.7B base.
The per-seat story is softening. The agentic story is compounding. The market is pricing the first and missing the second.
This is why your buyer thesis holds. The companies re-articulating their service around what the agent accomplishes rather than how many humans log in are the ones that get to grow through the repricing. The ones still selling seats are the ones getting cut in half. Salesforce down 26% YTD with "seat compression" in the earnings commentary is the counterexample that proves your point.
Really interesting take, especially the Global South angle, thanks for sharing. That feels underexplored. I do think the “anti” case is a bit underweighted though.
IMO, It’s not that enterprises fully replace SaaS. It’s that SaaS gets hollowed out. Systems of record stay, but get pushed down to the “database layer” while workflows move into agentic layers around them. That compresses value per seat without killing the product. The “companies won’t build software” point also feels like it’s shifting. AI is lowering the cost enough that anything close to revenue or operational leverage starts getting built internally.
On valuations, strong earnings today don’t really disprove the bear case. Feels like the market is repricing future margins and pricing power, not just current revenue. On the Global South angle, I wonder if this is insulation or just lag. The same forces making software more accessible could also make it more competitive and commoditized locally. The key question for me is how the The Innovator's Dilemma plays out here with incumbents.
Hard agree on build + buy though. That feels like the real shape of things.Curious how you think about where value actually accrues in this stack.
I do not disagree with the price pressure point - it needs to and will happen. But I think it will take a long time for enterprises to start building mission critical workflow on top of price-pressured systems of record. Why? maintenance. Building s/w is easy compared to maintaining it in bug free, cyber-safe stable-state forever. For small companies or non-mission critical things, sure maybe that will happen more quickly. But repalcement of enterprise grade saas systems by the enterprises that have been locked into a complex feature set for years will not happen anytime quickly regardless of the state of vibe coding, IMNSHO.
Thanks, really insightful, especially on the maintenance point, particularly as someone building in the space. Completely agree that’s where things get real and why core systems stick; nuance being less replacement, more workflow shift, so pressure shows up in value per seat rather than displacement.