AI's Impact on Startups and SaaS Draws Nuanced Debate
The a16z investor contrasts old startup rules with new AI-enabled possibilities.
Andrew Chen posted a comparison of startup conditions before and after AI. Before, hardware was hard, teams beat solo founders, only technical founders succeeded, PhDs built tech seeking problems, and hot SaaS firms commanded high multiples. After, hardware remains the sole AI-proof area, solo founders gain outsized leverage, anyone can build, and PhDs in AI infrastructure set their own pay. Separate posts note Meta's repeated internal chat tool switches and ongoing payments to vendors like Slack.
Combined views
596.9K
6 posts, first seen 4d ago
AI's Impact on Startups and SaaS Draws Nuanced Debate
The a16z investor contrasts old startup rules with new AI-enabled possibilities.
Andrew Chen posted a comparison of startup conditions before and after AI. Before, hardware was hard, teams beat solo founders, only technical founders succeeded, PhDs built tech seeking problems, and hot SaaS firms commanded high multiples. After, hardware remains the sole AI-proof area, solo founders gain outsized leverage, anyone can build, and PhDs in AI infrastructure set their own pay. Separate posts note Meta's repeated internal chat tool switches and ongoing payments to vendors like Slack.
Reactions
AC 
andrew chen@andrewchenBefore AI: - hardware is hard - teams > solo founders - technical founders only - PhDs build tech in search of a problem - hot SaaS startups get 100x ARR multiples After AI: - hardware is the only thing AI can't build - solo founders are 100x engineers - anyone can build - PhDs…
- likes: 1.1K
- replies: 169
- bookmarks: 492
- reposts: 84
GO Gergely Orosz@GergelyOroszThis was Meta They deprecated their in-house chat solution used a decade+ called Workchat less than a year ago. Decided to move to GChat. Now they’re moving again, GChat messages not being migrated Very strange there’s so little upfront planning, just YOLO then change course…
- likes: 848
- replies: 59
- bookmarks: 149
- reposts: 22
GO Gergely Orosz@GergelyOroszLook, good on Meta for just moving off now that it’s obvious that GChat so behind for agentic workflows. Eg Slack announced MCP support in Oct 2025, GA from this Feb - that’s 7 months GChat has dev preview for MCP today, only. Slack building so much more now…
- likes: 52
- replies: 2
- bookmarks: 3
- reposts: 1
GO Gergely Orosz@GergelyOroszTo everyone who said or liked posts on “SaaS is dead! AI is killing it” - no it’s not: Meta will be paying Slack easily ~$20M/year for a chat product Ask yourself: if they could build something with 2 devs + AI why not do it? They’d save SO MUCH. Because they clearly cannot!!…
- likes: 1.2K
- replies: 102
- bookmarks: 182
- reposts: 42
GO Gergely Orosz@GergelyOroszMeta has ~75,000 staff. They will spend hundreds of millions per year split across SaaS vendors. And they are using MORE vendors not less vs before! Eg they had an in-house chat product for ~10 years… then bought a vendor (because in-house sucked, even at Meta scale!)
- likes: 81
- replies: 5
- bookmarks: 6
- reposts: 1
JC John Crickett@johncrickett@GergelyOrosz Building a Slack clone is not their core business. They'd be better off buying the tool and using the "2 devs + AI" to generate new revenue streams. No good business is going to spend it's time replacing internal tools instead of creating better products.