bretpiatt

Total global wages ~$12 trillion/year. The "what you have to believe" here is that AI tool spend becomes at least 10-20% of it, my belief here is we see net slowdown in hiring as human + AI = super productivity so as businesses grow we'll continue to see increase in AI spending.

Then if inference does grow to a $2-3 trillion/year market that Anthropic is a big winner, as big as NVDA is now (they're going to be over $500 billion in revenue in calendar year 2027).

In the footer: Permit Pending is not a joke. Is this a joke to you?

Love it, I hope this becomes the most popular game on the internet so folks get to experience the Byzantine level of complexity we've made doing business in industries with physical stuff.

He initially stated in 2011: $30,000 in 2011 is worth $44,937.52 today.

New Model 3 today $36,990 so inflation adjusted already sub-$30k, and if Cybercab cones through even with inflation.

Available in agy for me as of now, I'm running my new model benmark of having it do comprehensive code audit on one of my more complex projects, it spun up 1 subagent per audit section and it took ~10 minutes per subagent on an 800k LOC, 60k complexity, ~4k file project. Will take me much longer to go through and determine the quality of the audit work.

    Antigravity CLI 1.3.4
    [My account email] (Google AI Ultra)
    Gemini 4 Argon (High)
    ~/[working folder]

  > /model
    ⎿  Model set to Gemini 4 Argon (High)

I don't know if we'll see price increases. My belief based on current market understanding is we'll see efficiency increases.

With models getting to the quality level they can perform a specific set of work at an acceptable quality level we will now begin to cost optimize that work.

We just invested in Lamb Labs from YC Summer 2026 as example company pioneering this next wave of ASICs to run a model.

We're going to see massively lower power usage and much better tokens/sec per dollar of capital expense as well. Think Moore's law but for tokens and output quality.

If agentic coding isn't dramatically changing your productivity you're getting left behind. There's a clear difference between vibe coding and AI assisted software / systems engineering from how I think about things. With AI you can be massively more productive. I do a lot of things beyond code and this is what I've built in 2026 (quick snip from scc, it has proper excludes for node_modules and other pulled in deps that I didn't build) from ~code/ on my dev box:

  ───────────────────────────────────────────────────────────────────────────────
  Language            Files       Lines    Blanks  Comments       Code Complexity
  ───────────────────────────────────────────────────────────────────────────────
  TypeScript         10,705   2,445,724   191,695    71,737  2,182,292    167,330
  JSON                1,435     604,218        58         0    604,160          0
  Rust                1,260     718,912    56,372    57,773    604,767     65,562
  Markdown            1,087     162,160    36,375         0    125,785          0
  SQL                 1,046     129,795     6,142     7,389    116,264      1,344
  Python                705     168,359    22,710    23,235    122,414     17,962
  JavaScript            657     209,397     8,824     7,587    192,986     45,383
  ───────────────────────────────────────────────────────────────────────────────
  Total              19,009   4,903,248   353,289   186,252  4,363,707    326,094
  ───────────────────────────────────────────────────────────────────────────────
  Estimated Cost to Develop (organic) $179,262,721
  Estimated Schedule Effort (organic) 98.79 months
  Estimated People Required (organic) 161.19
  ───────────────────────────────────────────────────────────────────────────────
  Processed 206,138,543 bytes, 206.139 megabytes (SI)
  ───────────────────────────────────────────────────────────────────────────────

I'm having a hard time understanding the value here based on these launch configurations and prices, what am I missing?

From what I'm experiencing, we have recursive self improvement that can find local maxima, I'm not seeing really strong evidence of unguided RSI that finds the true optimal.