Eric Friedman  ·  Tuesday, April 7, 2026

How Wide Is the
Silicon Valley
to Abundance?

AGI will produce extraordinary abundance. That is not the question. The question is the valley between here and there, and nobody knows how long it is.

A new format for me. Part essay, part visual exploration. I worked through a long Marc Andreessen tweet thread with an LLM, asked the questions I kept getting stuck on, and turned the answers into charts I could actually see. It all started with the tweet below.

Start scrolling to explore
It all started with this tweet

A long series of tweet essays from @pmarca on AGI, economics, and abundance. I thought they were fantastic. Worth reading in full before anything below.

Andreessen goes through a lot of scenarios, issues, and arguments. I wanted to feed the thread into an LLM, ask the questions I kept getting stuck on, and present some of what came back in a way I could actually understand. A lot of the text that follows is my framing of those exchanges. The visuals are where this clicked for me.

Section 01

The Argument

Marc Andreessen made a case worth engaging with seriously. It came out as a long series of tweet essays, and I thought they were fantastic. He walks through a stack of scenarios, counter-arguments, and historical analogies that most of the current AGI panic simply does not address.

What I wanted to do was feed the whole thing into an LLM, ask better questions about it, and present some of the results visually so I could sit with them. This interactive essay is what came out of that. Some of the text below is mine, some of it is my framing of the back and forth with a model. The charts are what helped it click.

The core of his argument: AGI unemployment panic is a 250-year-old logical error in better clothes. The Lump of Labor Fallacy, the assumption that work is a fixed pie. Machines take some, humans get less. Every generation of catastrophists has made this mistake. Every single one was wrong.

"The empirical track record of technological unemployment predictions is zero for approximately ten. Not mixed results. Zero for ten."
Marc Andreessen, 2026

The reason the fallacy fails: demand is not fixed. Work generates income. Income generates demand. Demand generates new categories of work. This is an engine, not a reservoir.

43%
Share of income Americans spent on food in 1900. Today it is ~10%. Agricultural mechanization freed that 33-point gap for industries that did not yet exist.
10/10
Times technology created new job categories at least as fast as it destroyed old ones. 250 years of industrialization. Base rate: 100%.
3.5%
US unemployment in 2019, the lowest peacetime rate in recorded history, after 30 years of computerization and internet saturation. The catastrophists were wrong again.

Four channels that break the fallacy

Channel 01
Productivity-Demand Loop
Automation lowers costs. Lower costs raise real income. Higher real income creates demand for new goods. New demand creates new jobs. The economy refills and exceeds its prior level.
Channel 02
The Baumol Effect
Easy-to-automate sectors get cheaper. Hard-to-automate sectors like care, craft, therapy, and performance get relatively more expensive and more in demand. AGI makes genuine human judgment more valuable, not less.
Channel 03
Comparative Advantage (Ricardo)
Even if AGI dominates humans at every task absolutely, humans retain comparative advantage somewhere. For human labor to have zero economic value, the math would need to be near-impossible.
Channel 04
New Industry Creation
Film editor, airline pilot, UX designer, podcast producer. None were predictable in 1900. The most important jobs from any tech revolution could not be foreseen. The base rate is 10 for 10.

What Andreessen grants: the transition will hurt. Real people. Real friction. Real communities that do not recover smoothly. Transitional friction is not structural permanence, but that distinction does not help the people living through it.

That gap between the theory and the people inside it is what I want to sit with.

Section 02

Rewriting the Hype Cycle

The Gartner Hype Cycle is one of the most useful mental models for where any technology sits. The beginning is great. The back half, the one with the trough, is where this piece lives.

My claim: we are approaching the peak of inflated expectations, not past it. The peak is right here in Artifact 1. The trough is still ahead. The distance between the peak of inflated expectations and the ramp of abundance is the trough of a valley that we are not sure exists, and if it does, we have no idea how wide it is. That valley is what this entire piece is about.

The Standard Gartner Model vs. The AGI Rewrite
Classic Gartner (reference)
AGI Hype Cycle
We are here
Time Expectations Peak Trough Plateau Peak of Inflated Exp. Trough Ramp of Abundance Innovation Trigger ← we are here approaching peak Artifact 1 — ericgfriedman.com

Past the trough, the road forks three ways. One is the path to abundance, which is what I hope for. One is a plateau of productivity, which would still be great. The third is a decline, and if it is really a decline, it probably means the end of a lot of what we take for granted. Artifact 2 is my attempt to draw those three outcomes on the same line.

One Line, Three Outcomes. The AGI Fork at the Trough
Shared path (trigger → peak → trough)
Valley — duration unknown
Optimist — up and to the right
Standard — plateau
Doomsday — down and to the right
Time → Expectations Innovation Trigger Peak of Inflated Expectations ? duration unknown we are here fork → ↗ Abundance → Plateau ↘ Decline Artifact 2 — ericgfriedman.com
Section 03

The Valley We Cannot Measure

This is happening now. There are real conversations in real companies, real roles being restructured, and real people who are going to lose their jobs because AI leverage means fewer people can do more work. The companies may perform better. The individuals affected do not feel better.

So the question is: how far and how long is the valley? That is why I put 1 to 25 years on Artifact 3. We don't know. We hope it's one year. Maybe it's ten. The idea of a temporary decline on the way to abundance is fine in the abstract. We can all paint a picture of the place of abundance I sketched in Artifact 2. What we don't know is how long it will take to get there, and a valley that is actually 25 years long is a very difficult valley to live through.

The macro outcome is not in question. The timeline is. The best performing VC portfolios look terrible before they look great.

"The valley is a feature of the cycle, not evidence the cycle is broken."
This holds for careers, companies, and economies alike.
The Valley We Cannot Measure — Between Peak Expectations and True Abundance
Time Prosperity / Employment Peak of Expectations AGI Abundance (higher + rising) we are here (approaching peak) 1 year? 25 years? we cannot know Artifact 3 — ericgfriedman.com

The valley we cannot measure is the biggest question in front of any of us. It could be one year, it could be five years, it could be ten years, it could be 25 years. Everyone sees a world where AGI abundance is amazing: robots that do the work we need, plentiful food resources, housing, and more. I'm leaving aside what will happen with money for the time being. It's the valley in between that no one understands right now. We're headed towards the peak of expectations, and over time we'll see how large that valley can be. Below, we start to explore which industries will have their own valleys.

Section 04

Every Industry Has Its Own Valley

Here is the idea I keep coming back to: the valley is not one thing. Every industry is going to have its own valley. Food has a displacement gap, where farmers lose their livelihoods before robot farming fills the shelves. Healthcare has a care access gap, where the strained system breaks further before AI-assisted medicine is really ready. Transportation has what I am calling the Mixed Road Danger Zone, the messy years when autonomous vehicles and human drivers share the road.

My question is not whether these things will improve. I believe they will. Everyone knows and hopes we are headed for autonomous driving, new medical breakthroughs, and abundant food. The question is the gap between here and there, and that gap is different in every industry. A different shape, a different depth, a different length.

The three charts below are not predictions of exact timing. They are a way to see that the valley splits into many valleys, and that each one will be navigated at its own pace by its own people.

Industry Valley: Food & Agriculture
Now 2030s 2040s 2050+ displacement gap Farmer economy food gap Robot farming surplus we are here Artifact 4 — ericgfriedman.com
Industry Valley: Healthcare & Medicine
Now 2030s 2040s 2050+ care access gap Strained system care gap AI surgery + longevity we are here Artifact 5 — ericgfriedman.com
Industry Valley: Transportation & Autonomous Vehicles
Now 2030s 2040s 2050+ mixed-road danger zone Human drivers driver job loss Zero accidents + free freight we are here Artifact 6 — ericgfriedman.com

The key question is not whether we get to abundance in each sector. It is: who bears the cost of the crossing, and for how long? The farmer whose business collapses before robot agriculture scales does not care that the macro outcome is positive. The nurse displaced before AI-assisted medicine is ready does not feel the long-run numbers.

This is the part of Andreessen's argument I find incomplete. He is right about the destination. He is less specific about the people inside the valley.

Section 05

Regenerative Abundance

With enough time and compounding, we reach a place where the basic structure of human need gets rewritten. Robot labor addresses food and shelter at scale. Self-replicating systems lower the marginal cost of production toward zero. The bottom tiers of Maslow's pyramid, the ones all of human history has fought over, become table stakes.

What remains stubbornly human: belonging, meaning, purpose. A lot of people are already talking about these. The one I think is missing from most of this conversation is taste. The ability to sense what is good, what is worth making, what is worth spending a human life on. Taste is not replaced by AGI. It gets more valuable, not less.

Artifact 7 is my attempt to line these up against the tiers and timelines, so we can see where abundance arrives first, where it arrives later, and where it never really arrives at all.

Maslow's Hierarchy vs. AGI Progress — What Gets Solved and When
Physiological: Food, Water, Shelter Robot labor · abundance economics · automated supply chains Safety: Security, Health, Stability AGI diagnostics · autonomous infrastructure · predictive systems Love and Belonging: Community, Connection Assisted by AI · not replaceable by it Esteem: Respect, Achievement, Purpose Redefined by AGI · not replaced by it Self-Actualization Meaning · creativity · love Irreducibly human Now –10yr 10– 25yr 25+ yr ? Always Artifact 7 — ericgfriedman.com
Section 06

The 3 Paths

Everyone wants the hockey stick. Most people are in the loop. The staircase is underrated. All three, if you stay with them long enough, arrive at roughly the same destination. The fidelity of the journey is what differs.

The hockey stick is the 10-year overnight success. Up and to the right. It is what people see in the highlight reel: the individual who made it, the company that broke out, the musician who suddenly got discovered, the artist who finally got the show. It is real, and it is rare. Most people and most companies are not on it.

The loop-de-loop is what most real trajectories actually look like. The top of the loop is when things are going great and you get too close to the sun. You get fired, laid off, your round falls apart, the launch does not land. The loop is usually a retread. For a lot of people in the valley we just drew, the loop is going to be the loss of a job, which is also a loss of meaning and a loss of skill. The ride back up is retraining, rediscovery, and finding a new purpose. That fits a loop, not a line.

I wanted to show this for both the personal path and the company path, because I have watched both happen in startups and in careers. The company chart under the Company tab is the same shape for roughly the same reasons.

Hockey Stick — the 10-year overnight success, near-vertical at the end
Staircase — sustained work, then a vertical jump, then work again
Loop-de-loop — full loops: layoffs, failures, comebacks. Same destination.
Time (career arc) Success / Impact Goal Hockey stick grind promotion new role breakout Crash 1 layoff comeback Crash 2 pivot/fail comeback same destination Artifact 7 — ericgfriedman.com
Hockey Stick — the unicorn narrative, near-vertical at exit
Staircase — pre-PMF, PMF, growth, scale, exit
Loop-de-loop — pivots, near-death moments, comebacks
Time (company lifecycle) Revenue / Value Exit Hockey stick pre-PMF PMF growth scale Crash 1 pivot/near-death comeback Crash 2 pivot comeback same destination Artifact 8 — ericgfriedman.com

The loop-de-loop is not failure. It is the highest-fidelity path. It goes higher first, and it crashes the hardest. But it arrives at the same destination as the staircase. The people who navigated it learned things the staircase never teaches, because you cannot read about near-death. You have to survive it.

Section 07

What I Think

I am an optimist, but I am also a realist. The path to abundance is real, and I think it will be extraordinary when we get there. What I don't know is how big the valley is going to be. I think Andreessen is right on the macro, and I think the shape of the cycle plays out roughly the way he describes.

The economy is an engine, not a reservoir. It has refilled every single time. The AGI era will be no different in structure, only in speed, which is the part that actually keeps me up at night.

The thing worth doing is saying the hard part out loud. Name the valley. Identify it clearly. Acknowledge it in every industry. See what is coming, understand what the loops are, what the staircases are, and where we are actually trying to get to as a human race. We all want to be up and to the right. What we are really trying to figure out is how long the valley is before we get there.

These are the questions I keep turning over, for myself and for the companies I spend time with:

I welcome answers to these questions, new questions, or continuing the discussion on X.

@ericfriedman

Reply directly to the announcement