top of page
Search

I Still Hate SaaS (Subscription Software)

How $49.99 Software Turned into $1,149/Year — And Why AI Agents Are Making It Obsolete

I’ve hated SaaS subscriptions for years. Not because I’m stuck in the past — I follow tech trends closely for my clients. But the switch from buying software once to paying forever felt like a massive wealth transfer from users to vendors, with little real added value in return.

What we got? Often slower performance, more friction, and bills that never stop rising.

SaaS = Software as a Service

My gut wasn’t wrong. Markets are now catching up. SaaS multiples are compressing hard, while companies delivering genuine breakthroughs — AI chips, autonomy, energy, defense tech — command sky-high valuations. AI agents are the big reveal: a lot of what these subscriptions charge for is now easy (and faster) to do yourself, locally, without the rent.


Exhibit A: This box.


QuickBooks Simple Start 2008. One-time price: $49.99. You owned it forever. No monthly dings, no forced "upgrades," no waiting for fields to load. Type at 80 wpm, tab through pages instantly — pure, snappy efficiency.


In 2026? QuickBooks Desktop Pro Plus renewals hit $1,149 per year for one user (up from $999 effective Feb 1, 2026). Extra users? $230 each. Online Plus lists at $115/month ($1,380/year) once promos end. Intuit’s pro tax tools? Pay-per-return still ~$65 federal + $57 state — often netting Intuit more per return than the preparer keeps.


That’s not progress. That’s extraction.


Salesforce? Core is 1990s tech: big database in spreadsheet clothing. Enterprise edition: $175/user/month. 40 users = ~$84,000/year locked in forever.


Adobe? Photoshop CS6 perpetual was $700–$900 one-time back then. Creative Cloud Pro now $69.99/month ($840/year) forever — subscription-only, no escape.


Microsoft Azure/365? Forced recurring charges replaces old one-time Office buys, with usage billing that surprises firms with big runaway tabs.


Here’s the stark comparison:

Software

Then (One-Time or Low Annual)

Now (2026 Recurring)

Multiple Increase

Key Friction Added

QuickBooks Simple Start

$49.99 one-time (2008)

$1,149/year Desktop Pro Plus

~23x

Latency, no ownership

QuickBooks Online Plus

N/A

$115/mo ($1,380/year)

N/A

Visible lag even on fiber

Salesforce Enterprise

N/A (cheap perpetual alternatives)

$175/user/mo (~$84k/year for 40 users)

Massive recurring

Lock-in, slow UI

Adobe Creative Cloud

~$700–$900 perpetual (CS6 era)

$69.99/mo (~$840/year)

~10–20x over time

Subscription-only

Microsoft 365/Azure

One-time Office ~$250–$400

$12.50+/user/mo + usage fees

Recurring + surprises

Cloud latency, surprise bills

Even on the fastest fiber I can buy from Spectrum, cloud tools lag. Desktop software lets me fly through fields at full typing speed. Browser-based? Tiny but constant delays — network round trips, HTML rendering, JavaScript bloat. Over a workday, it kills efficiency. Cloud + infinite scale should have slashed costs, not jacked them 5–20x.


SaaS added patches, real-time collab, mobile. Scaling teams benefit. But the core — basic workflows, spreadsheet databases — is ancient tech. It should cost pennies compared to 1999, not fortunes.


Most “AI” add-ons? Lame. I’ve been pitched 20+ AI notetakers/schedulers — glorified search + summary. Even RIA tools like Altruist’s Hazel feel like fancy note/search layers, not worth thousands/year. Real change (agentic AI change) isn’t bolt-ons; it’s full replacement.


Gartner says 40% of enterprise apps will have task-specific AI agents by end-2026 (from <5% now). By 2030, agents could absorb/replace 35%+ of point SaaS. Agents do entire flows: reconcile books, enter CRM data/follow up, log meetings to actions, model tax scenarios — locally or open-source.


Today’s no-lock-in options:

  • Accounting: Akaunting or GnuCash (free, self-hosted) + local LLM (like Ollama) for smart reports.

  • CRM: Twenty (open-source Salesforce alternative), Odoo Community, SuiteCRM.

  • Creative: GIMP/Inkscape + local image gen (Stable Diffusion/Flux) + n8n workflows.

  • Full private stack: Supabase database + n8n automation + Grok/Claude/local model = your own fast, private “QuickBooks + Salesforce + tax” agent.


Zero latency, full speed, your data private and on your machine.


Markets price true value: Higher P/E in AI innovators signals real value creation per share (exponential productivity). Compressed P/E in legacy SaaS? Rent extraction finally exposed.


Current multiples (Feb 2026):

  • Intuit (INTU): Trailing P/E ~24–26 (down from 48–60+ peaks).

  • Salesforce (CRM): ~23–25.

  • Microsoft (MSFT): ~24–25.

  • Nvidia (NVDA): ~47.

  • Tesla (TSLA): ~370+.

  • Anduril (private defense/AI): In talks for $60B+ valuation (up from $30.5B last year; secondary market implying even higher like $80B+). No public P/E (private), but 68x revenue multiple on ~$1B 2024 sales screams premium for breakthroughs. Founded by Palmer Luckey — the kid who invented Oculus in his garage, sold to Facebook for billions, got fired after supporting Trump, won a huge wrongful termination settlement ($100M+ reported), and now literally builds "The Borg": swarms of autonomous drones and submarines that think and act as one. That’s innovation.


Post-2008 low rates primed the pump; post-COVID lockdowns + stimulus + ZIRP supercharged the bubble. Now rates normalize, cash yields real returns (~3.5–4% T-bills), and capital flees these SaaS tollbooths for deflationary builders.


For RIA clients and practices: Audit your SaaS bleed. Redirect savings to portfolios or growth — it compounds. Spot the reallocation: legacy SaaS compresses, real innovators expand. Clients who act early outperform.


My hatred for SaaS wasn’t just emotion. It was pattern recognition. AI agents made the math undeniable.


Download my free SaaS Audit Checklist below — or drop your biggest subscription headache in the comments. I read every one.




 
 
 

Comments

bottom of page