The salary inflation in AI is getting ridiculous.

I went through this exercise myself when someone on a forum claimed "everyone I know in AI makes $1M+." It's junk. Here's what the data actually says:

• Top advertised base+bonus+equity: $824k (a senior research role at one of the big labs)
• Typical "elite" AI engineer, 5+ years: $350k – $550k total comp
• New grad AI/ML roles: $150k – $220k at top firms
• The $100M claim: usually founders or early equity holders, not hired employees

Why does this gap matter? Because the $100M narrative distorts career decisions. People chase "AI exit" dreams and pass on solid, well-paid roles that are actually attainable. The $824k job exists — it's a real posting, a real team, real responsibilities. But it's one position, and it's still a W-2, not a lottery ticket.

The closest real-world path to those 8-figure numbers is:

1. Join a pre-IPO lab or startup at the founding-engineer level. Equity, not salary, is the multiplier.
2. Build something demonstrably useful — a deployed agent, an eval harness, a fine-tuning pipeline that ships. Papers alone don't get you the 7-figure packages anymore.
3. Be willing to relocate or go remote-first. The $824k role wasn't in San Francisco; it was in a market with less talent density.

So yes, AI is the best-paid engineering niche right now. But the useful question isn't "how big can the number get" — it's "what compensation is actually being offered today." $824k is a great answer. Everything above that is salesmanship.