Finding signal on Twitter is more difficult than it used to be. We curate the best tweets on topics like AI, startups, and product development every weekday so you can focus on what matters.
Everyone claims that SaaS is dead. They're all wrong. | Tech Twitter
Everyone claims that SaaS is dead. They're all wrong.
Tyler Denk@denk_tweets · September 1, 2026 · 8 min read
Originally published by @denk_tweets on X. Tech Twitter preserves the original source alongside this readable edition.
Before you dive in
âą The post demonstrates that despite AI making custom software development technically feasible, SaaS remains economically rational because the hidden costs of building and maintaining bespoke solutionsâincluding infrastructure, integrations, ongoing maintenance, and opportunity costsâconsistently exceed subscription fees, especially for non-technical businesses
âą For builders and founders, this means the SaaSpocalypse narrative is overblown; the real opportunity lies in improving SaaS products to justify their value proposition rather than betting on mass defection to custom-built alternatives, while AI practitioners should focus on enhancing existing platforms rather than assuming they'll become obsolete.
Best for builders who want practical takeaways. 8 min read.
You may have heard of the SaaSpocalypse.
The thesis: with the rise of AI agents and coding tools, people and businesses are all going to build their own bespoke software, leaving these incumbent software businesses behind with an ever-diminishing customer base. That, and the rise of agentic workflows that will destroy the per-seat model that has grown so popular these past few decades.
Every thinkboi on X shares this vision, claiming that software as we know it is dead. And in the past year, the SaaSpocalypse has gone from fringe to consensus.
The markets have taken note. Shares in public SaaS companies have gotten crushed, despite many of those businesses performing arguably better than ever.
Letâs look at HubSpot, for example. Over eight quarters HubSpot compounded revenue at ~20%, expanded margins nearly every quarter, went GAAP-profitable, and grew free cash flow more than 50% â hitting its 2027 margin target a year ahead of schedule.
But the stock is down ~55% from its 52-week high (bottomed out at 68%).
The market hates uncertainty. Sure, performance has been good (arguably excellent), but the acceleration of AI is so âunprecedentedâ that itâs impossible to bet on the status quo (i.e. HubSpot remaining relevant in an AI-first world).
Betting on the SaaSpocalypse means betting on the thesis that all of these companies will soon be moot in a world where AI deploys custom software for everyone. That every SMB on earth is just one Claude Code session away from firing their entire stack, and that nobody will ever have to maintain any of it.
Iâm not betting against Anthropic (or AI), but I am here to tell you that the SaaSpocalypse is totally overblown and complete bullshit.
Most of those claiming that SaaS is dead have never actually built anything at all. They arenât in the arena using the latest tools. Theyâve never built or run a business at scale. They probably think that Opus is some sort of peptide.
Let me tell you why theyâre all wrong. Actually, let me take it from the very top.
Here are a few examples of SaaS companies: Shopify, Adobe, Zoom, Canva, Salesforce, beehiiv, etc. Software as a service (SaaS) is exactly what it sounds like â you pay someone to provide a service you don't have to build yourself.
When you host your storefront on Shopify, you donât need to build your own payment infrastructure, or host your own website, or think about sales tax across all 50 states, VAT in the EU, PCI compliance, chargeback disputes, or fraud.
We doomscroll, you upskill
Get the 10 tweets shaping how builders think today.
Newsletter
We doomscroll, you upskill
Get the 10 tweets shaping how builders think today.
In fact, they have a team of nearly 8,000 people who specialize in e-commerce, payments, checkout, fraud, logistics, and everything else to ensure your storefront is best positioned for success. They even routinely push updates to make the Shopify software better while you sleep.
You get best-in-class infrastructure for your online storefront, and no one on your team even thinks about it at all. From that perspective, the $39 per month youâre paying Shopify feels like a steal. Thatâs the beauty of SaaS.
We can use beehiiv as an example too. Technically, anyone has been able to build their own email infrastructure for decades. The same way anyone could build their own storefront for e-commerce. AI just makes it easier, and puts it within reach of people who aren't technical.
With the help of AI, sure, you could spend thousands of dollars and several months building your own email platform and have your own (much worse) version of beehiiv, custom-built just for you.
But when your team needs to integrate the newsletter with your podcast, or send dynamic content to your readers, or have access to enriched audience data⊠youâll need to go back to the well and have someone on your team build these features themselves.
When you want to monetize your newsletter, whether with ads or subscriptions, youâll need to build your own sales pipeline, reporting, user authentication, payment rails, renewal logic, customer alerts, and more.
When your emails start to land in spam, your team is now solely responsible for figuring out email deliverability all on their own.
And remember â replacing SaaS and building the software yourself isnât free. You still need to pay for the infrastructure, servers, maintenance, monitoring, and the other software and plugins required to make it all work. Not to mention the added human capital required to build and manage everything (indefinitely).
So after all of the time, money, and labor required to build a worse version of beehiiv, youâre now stuck having to maintain and update it all on your own forever. Or, hear me out, you could pay a small premium to have 130 specialists own all of that for you. That premium comes with regular software updates, access to the latest tools, uptime guarantees, and a team of experts in email deliverability and creator tools who can help you whenever you need it.
Thatâs the crux of the SaaSpocalypse debate. And thereâs a small vocal minority of indie hackers who run solo teams claiming that everyone is going to build their own software just because itâs now technically possible.
If that were true, you would expect me (the CEO of a tech company, with 50+ engineers and a huge bias towards building vs buying) to have totally replaced all of our SaaS spend. But we still happily pay for Slack, Sentry, Linear, HubSpot, Loom, Zendesk, Amplitude, Ramp, Notion, Customer.io, etc. In fact, we pay considerably more for these today than we did just a year ago.
As my COO, Dan, says: weâve been able to bake bread forever. Most people still buy it.
Meanwhile, Iâm not just yapping from the peanut gallery talking hypotheticals either. Iâm in the arena. Iâve done the work myself.
Last week, I announced that a few friends and I launched a social wellness club in MedellĂn. Itâs called MADRE, and itâs home to the largest sauna in all of Latin America. (By the way, if you live there you should book a session. And if you donât live in MedellĂn, you should follow us on Instagram).
We could have used Mindbody, WellnessLiving, Walla, or any of the other platforms to manage bookings and memberships. It would have cost less than $100 a month and been ready to use in a few hours. SaaS baby.
But out of curiosity, watching this whole vibe coding movement from the sidelines on X⊠I wanted to take advantage of the opportunity to get my hands dirty and build it myself.
As for my credentials: I'm a self-taught developer. I led product and engineering at Morning Brew, was a Product Lead at YouTube, and currently run product at beehiiv. I've built a lot of software, and I think I have a decent eye for design.
Hereâs the MADRE landing page to vouch for that design eye đ
Keep all of that in mind as I explain what happened nextâŠ
I spent about 6 full weekends, Iâd estimate maybe 80+ hours, building this website and platform from scratch. I probably spent another 20+ hours doing QA to ensure everything works perfectly.
I had to build custom integrations with Twilio, Resend, Customer.io, Vercel, Slack, Sentry, Supabase, and Wompi (a local payment provider in Colombia). Yes, pay for and configure each of those individual accounts as well. I built an admin portal so our managers can run the business. I had to purchase datĂĄfonos (payment terminals) to build and test the physical payment infrastructure.
I also had to create user guides and tutorials to hand off to the team so they could learn and operate the software on their own.
And thatâs just the upfront investment. When the staff provided feedback, I had to build new features to support their requests. I also have Sentry errors piping into Slack to alert the team whenever thereâs an issue with the platform.
As for who has to update and fix the software when itâs broken? Thatâs on me. But I have a full-time job running beehiiv and donât have the bandwidth to do that outside a couple of hours here and there on the weekends (uh oh).
Add up Twilio, Vercel, Customer.io, Sentry, Resend, and Supabase and we're paying multiples more than what a SaaS company would charge to handle all of it for us. And that SaaS company would be improving the product, watching it for problems, and answering the phone when something breaks.
For the low, low price of $300 per month, and hours of my life I'll never get back⊠we have a fully custom-built solution that we can edit and tweak however we'd like :).
Remember my credentials? I've led product and engineering at multiple startups and it still took me a hundred hours. Hell, Iâm still technically on-call to support it.
Do you think the average person selling cowboy boots in Jackson Hole is going to cancel their $39 per month Shopify subscription to spend 100+ hours building their own store, and then hire someone to manage it forever?
Of course not. And the millions of small businesses paying for Salesforce arenât going to build their own CRMs either.
