Marc Campagna

[Gokul Rajaram @gokulr]

THE TOKEN HANGOVER

@matanSF (Matan Grinberg), CEO and co-founder of @FactoryAI , interviewed by @HarryStebbings (@20vcFund )

This is a special for me since I've been an investor in @FactoryAI since their seed round, and think Matan is a very very special founder.

Summary: Grinberg argues the next 24 months in enterprise AI are a resource-allocation problem: tokens, dollars, and people. Most CIOs are now waking up to bills they cannot justify. The fix is to spend frontier tokens only on the 10-20% of work that requires planning intelligence, run the other 80-90% on open models, and rebuild teams around load-bearing polymaths who own business outcomes. The single-frontier-monopoly fear is fading: four roughly-equivalent labs is the emerging reality, which puts pricing power back in the application layer.

1. The Token Hangover. Enterprise AI adoption ran through three phases this year: boards yelling at CEOs about AI strategy, "token maxing" with AI usage written into perf reviews, and now the morning-after bill. One CIO Grinberg spoke to was spending hundreds of thousands of dollars a month on engineers asking Opus 4.8 things like "how's it going" and "what are my macros from lunch." The frontier model became the default surface for every question, no matter how trivial. Phase 3 is the moment routing matters: every call to a frontier model needs to earn its price.

2. Resource Allocation Is the Job. For the next 24 months every C-suite is solving the same problem: how to allocate dollars, tokens, and headcount against business outcomes. Engineering teams used to be judged by features shipped per quarter, a metric with no link to revenue, market share, or retention. A logistics company adding more engineers to ship more features was always solving the wrong problem; AI made the misallocation visible. Tie every person's work to the metric that actually moves the business, then re-allocate.

3. Load-Bearing Individuals. The "10x engineer" frame measures lines of code, the wrong unit. Grinberg's unit is the load-bearing individual: the person whose absence breaks something. With AI the load-bearing few compound roughly 10,000%; the others get close to nothing, so any org enforcing one token-spend-per-engineer number is painting with too wide a brush. Average token spend per engineer will land on the same order of magnitude as their salary within three years, with a wildly bimodal distribution.

4. Frontier for Decisions Only. 80-90% of software development tasks can run on open models; the remaining 10-20% is planning, where the frontier still wins. This mirrors how human orgs work: leadership is a tiny share of total hours but decides the company's fate. The ego trap is engineers assuming their work is too important for an open model. The router decides better than the engineer, and the cost curve falls only if you wire the routing.

5. The Kirkland Mistake. Kirkland & Ellis announced a $500M, five-year internal AI build, which Grinberg reads as validation for Harvey rather than a threat. Building AI is not a law firm's core competency, and Kirkland's spend will teach them how hard it is. The general rule: just because you can build it does not mean you should, and the discipline is naming the few things you and your team own end-to-end. Outsource everything else, even when you technically know how to do it yourself.

6. Model-App Separation. When the model provider also sells the app, the incentives split: an API business wants you to spend more tokens. A healthy market keeps the application layer independent, so model providers compete on price, speed, and quality every week. Enterprises do not want to vendor-lock again; every CIO carries scars from the cloud era's three-year discount-then-jack-the-price trap. The application layer survives precisely because it forces that competition.

7. Sales as Product. Name a legendary company with a weak sales or marketing team. You can't. The Silicon Valley fallacy that research sits at the top and sales is "dirty work" produces companies that win the gold rush and then collapse when gravity returns. At Factory, engineers and salespeople sit intermixed; when sales closes, engineering says "we closed"; when engineering ships, sales says "we shipped." Atrophied sales muscles will not regrow once enterprise buyers stop saying yes to everything.

8. Polymath Era. Da Vinci, Newton, Euler could be polymaths because their fields were shallow. By the 2010s a theoretical physicist needed 50 years to reach the frontier before contributing anything new. AI collapses that catch-up time, so one person can push forward developer marketing, token-caching infrastructure, and solution engineering at once. The engineer of the future is a GM who owns marketing copy, product metrics, and sales enablement.

9. Build the Factory. Factory's name is literal: engineers in the next era design the assembly line that produces software. The DevX investments that used to scale linearly with headcount (good docs, CI/CD, linters, pre-commit hooks) now scale with the number of agents you run, which is 10x or 100x larger. Every dollar spent making agents production-ready compounds against thousands of PRs a week. Humans move up the stack, from writing code to designing the system that writes code.

10. Seal Team Six. Mandating beds in the office is a hiring failure dressed up as commitment. Grinberg's image: a basketball game judged by who sweat the most, when the scoreboard is what counts. Factory bought eight sleeps for all 30 team members at the time, because recovery is where the gains come from when work requires every ounce of brain power. If your load-bearing engineer can do their best work on two hours of sleep, they were not doing load-bearing work in the first place.

11. Four Frontier Labs. Grinberg's biggest mind-change this year: a single dominant model is unlikely, and four roughly-equivalent frontier providers is the more probable steady state. That outcome is the win for humanity. A one-lab monopoly was the dangerous scenario, and four equivalent labs is also the structural bull case for the application layer because it forces real ongoing price competition. Every CIO Grinberg meets has already decided not to throw their lot in with a single provider.

12. Dario's Self-Serving Doom. "AI will take your jobs" was the pitch that helped raise hundreds of billions, and Grinberg thinks it damaged public psychology and fed the slow-AI lobby. Watch the rhetoric flip at IPO: humans will suddenly become important again, because humans are the ones buying the stock. Founders who never needed to raise that money, like Zuckerberg and Hassabis, never made that argument. Incentives drive the labor-displacement rhetoric more than philosophy does.

[GEOFF @geoffreywoo]

The best diligence question is usually not "how big can this get?"

It is "what has to be painfully true for this tiny wedge to keep expanding?"

That question exposes the actual company.

[Aaron Levie @levie]

Good thought provoking post from Anthropic. I think this paragraph points to the key element of the optimistic scenario of AI:

“There has been an explosion of new ideas, initiatives, tools, and simulations, as a result of Anthropic employees working with highly capable models—far more than we have the capacity to pursue. The rate at which organizations can spot and fix these bottlenecks may be a skill that improves over time, and it may become the most important skill for any organization.”

AI lowers the barrier dramatically to allowing us to do more. As a result of that, we have far more ideas than we can pursue, and for the ones that we want to pursue we’re ultimately limited by our ability to go take on the surrounding work to execute those ideas. There’s almost no amount of AI progress that can happen where that goes away.

AI is going to let us build much more software, launch more marketing campaigns, research more drugs, and so on. All of this work, even when augmented by agents, still ultimately requires people to manage.

[Elad Gil @eladgil]

The events of the last 6 months in technology are arguable amongst the most important in human history

The tools now increasingly exist for recursive self improvement of models & agents

We are likely in very early lift off & exponential

Largely unnoticed outside of tech

[Marc Campagna @markymarkymarcc]

@kazdenc missing @gaiia_software reppin' Quebec City

[A Deal is Imminent @tradeoilstocks]

“Give them a massive amount of oil, agricultural land, copper, freshwater, and every natural resource in the world. Now make them neighbors with the biggest market in the world. Great, now have them leave the resources in the ground and instead flip condos to each other”.

[Joshua Kushner @JoshuaKushner]

what if everything goes right

[tobi lutke @tobi]

Also don’t remember. But it’s been living rent free in my head ever since. Puts a very important truth that you intuit when building companies in a very accessible form.

[Scott Stevenson @scottastevenson]

One of the bigger meta-patterns I've noticed is that engineers and conscientious people tend to overweight the value of internal consistency and logical consistency

Our audiences are barely paying attention, and it is more important to resonate in simple ways than to worry deeply about the precision and consistency of our systems and logic

Reality is incredibly complex, and any illusion you have that you have figured out a "consistent logical formula" for your work is probably wrong and unimportant

Vibes matter a lot more than people think in a hyperdimensional world

[Hubert Thieblot @hthieblot]

You became a founder.

You quit the 9-to-5. You raised a little money. Everyone called you "brave" over drinks. You spent your nights building and your days pitching "the future," convinced that the next launch would change your life forever.

Then a year passes.

Flatline traction. $0 salary. Your co-founder quit via Slack. Your girlfriend left for someone with a 401k and a "stable" future. Your friends are posting house keys while you’re staring at a bowl of ramen, rehearsing the same tired lies to your parents about why the "big break" is just around the corner.

Is this the end? You start wondering if you made a mistake.

No. You keep telling yourself every founders went through this at some points. But you don’t stop.

Logic says quit. Your ego says run. But there’s a sickness in you that won't let go. You’d rather fail at this than succeed at anything else. You tell yourself it’s just one more launch, one more pivot, one more "yes."

You’re not delusional; you’re committed.

You’ll miss this.

Not the stress, but the electricity. The raw doubt that forced you to grow. The quiet fire of building while the world slept. The pure, unrefined dopamine of that very first user.

These aren't just "hard years", they are the years that forge you. One day, when the bank account is full but the mystery is gone, you’ll find yourself wishing you could feel this hungry again.

They all do.

[Hubert Thieblot @hthieblot]

Never quit as a founder. I’m begging you.

It’s 0 for longer than you’ll ever expect. No momentum. Soul-crushing doubts. Nobody seems to care. Even when it looks like it’s working, it’s not. You keep trying new things. You don’t lose hope.

Then it snaps to 100. You finally find the one thing that resonates. You wake up with more customers than you can handle. Everything is breaking. Momentum keeps building even when you’re not pushing. Something changed.

You didn’t get lucky, you just didn’t leave.

[Marc Andreessen 🇺🇸 @pmarca]

"Surprisingly, occupations with higher exposure to AI have grown faster than least-exposed ones, not slower."

Not surprising! Productivity growth -> economic growth -> job growth.

[HotSotin 🇫🇮🇺🇦🇪🇺△ @HotSotin]

"The evils of capitalism:"

[Joshua Kushner @JoshuaKushner]

C.S. Lewis: The Inner Ring

[Gokul Rajaram @gokulr]

EQUITY VESTING

I’m starting to see more founders put into place 6 year vests for founding / early employees, and I think it really aligns incentives around thinking long term and sticking around to build an enduring company.

Founders: don’t take 4 year vests as written in stone. Rethink it from first principles, considering what you’re trying to build.

4 years made sense in an era when the average time to IPO was 4-5 years. That era is gone. The best companies today take 10-15 years to reach their full potential. If your vest schedule is shorter than your ambition, you’ve created a misalignment.

A few things worth rethinking:
•Cliff length (discussed above)
•Back-weighted schedules (more equity in years 4-6, not front-loaded; Amazon has done this for many years?
•Refresh cadence (how do you keep people incentivized post-vest without diluting everyone?)

The vest schedule is a cultural document. It signals how long you think this will take, and how serious you are about people staying for the whole ride.

[Brett Berson @brettberson]

Being a Chief People Officer is not for the faint of heart...

[sam @samuelchenard]

we kept building agents for our startup and hitting the same wall — giving them email just sucked. your inbox is a no-go, gmail needs a human to set up, and every other solution still requires you to sign up on behalf of the agent.
so we shipped lobstermail. the agent signs itself up, no human in the loop.

[Gokul Rajaram @gokulr]

Speaking with Head of Product at a large company (30+ PMs). They actively ship code and their expectation is that all PMs on their team do this.

Their recommendation for the first 3 things that PMs and designers (and marketing!) should own end to end:

1. Public-facing website

2. Support experience

3. Onboarding / growth flows

[Devon Eriksen @Devon_Eriksen_]

Many people, even self-described conservatives, think socialism would work if human nature were different.

No. Socialism cannot work, even in a hypothetical society of selfless genius saints.

Why not?

Because socialism centralizes economic choices. How much lumber do we produce? How much wheat? What should the hourly wage of a garbage collector be? How much should insulin cost? How about bread?

Socialists think that if you elect the right people, they will make these decisions intelligently and altruistically, and everything will be great.

But it doesn't matter how smart and benevolent you are... you can't make a good decision without the right information. The Socialist Central Planning Committee, however wise or benevolent, doesn't know what's wanted, or what's available, because that information is conveyed in prices, and accurate pricing is the very thing that socialist governments wipe away with the bureaucratic pen.

Capitalist networks are decentralized. They distribute decision making to where the information is.

A man selling metal doesn't know anything about desks, or lumber. He doesn't know how many desks people want, or whether they should be made out of oak, or folded metal.

But he does know how much it costs him to smelt iron ore into steel, and roll it into sheets. So he sets a price, and others decide whether, and how much, to buy.

That price contains the information others need to decide whether steel is plentiful, and should be folded into anything you can make out of sheet metal, or is scarce, and should be saved for things that can only be done with steel, and furniture should be made out of oak, or pine, instead.

Socialism works, or rather doesn't, by using the threat of force to set the prices of things, or take money from one person and give it to another.

But every time this happens, critical data on supply or demand is erased... data that you need to make decisions.

Individual prices are a decision, a guess at where supply and demand cross paths. But since free markets reward those who guess correctly, or copy a correct guess, aggregate prices are data on supply and demand.

For a socialist central planning committee to order the manufacture of the correct number of cars, or to correctly set the price of a car, they need to know a thousand thousand thousand things about steel and aluminium, welders and assembly robots, rubber and glass and lithium batteries and copper wire, which they must gather, along with trillions of other pieces of data, from literally everyone in their entire civilization.

Tesla only needs to know how much people charge them for the stuff they need.

At every transaction in a captialist society, vital data is compressed into its most compact and useful form, then passed along to the adjacent step, where abundant brainpower is waiting to make decisions with it.

Any defective node in the web that fails to make good decisions receives swift and automatic feedback, and either heeds that feedback or goes out of business, to be replaced by someone who will.

Yes, in a capitalist system, there are many undesirable results. But capitalism doesn't create these results. It discovers them. They are inevitable consequences of the state of technology, and will persist until something is invented that changes the terrain.

In socialism, no such solution is possible, because all the inherent problems you need to solve with progress are hidden from view by the far worse problems you created for yourself by separating the place where decisions are made from the place where information is known.

[Ian Cassel @iancassel]

So good by @GavinSBaker

In Michael Jordan’s second to last year with the Chicago Bulls, Jerry Krause, GM of the Bulls said, “Listen players don’t win championships, organizations do. It isn’t just Jordan or Pippen. It’s how we scout, draft, train, it’s our system." Well… after Michael Jordan left they never won another championship.

The thing allocators struggle with is it feels safe and good to say there is a process and its repeatable. Yes, it is important to have a process that is repeatable that works for you, but any process that is repeatable that generates significant alpha will be quickly arbed away in a competitive world.

So where does repeatable outperformance come from? I would submit that any investment organization no matter how big, there are somewhere between 2-10 people where if you took those people out, and that organization would have the exact same process, the results would be dramatically different.

https://t.co/YEfcvZQmKN