First company sold for nearly a billion dollars. Second raised about seventy-five million, then shut down in under three years.

Between them sat Twitch’s sale to Amazon and Atrium—a star startup that wanted to “fix legal” with software plus a law firm. Investors included Andreessen Horowitz. When TechCrunch covered the shutdown, the figure was about $75.5 million raised, just over a hundred remaining staff laid off, and some leftover capital returned to investors.

Silicon Valley loves the story that enough money keeps the ship through the storm. Atrium proved the opposite:

Some holes money cannot plug. Some companies are more honest closed early than kept on life support.

Money that arrives too fast looks like victory. It is an accelerator.

Around 2014, Twitch—grown out of Justin.tv—sold to Amazon for a reported ~$970 million. Kan had been a YC partner; trust was not scarce when he started again. He later admitted the ego was still hungry for the biggest possible company—dreams full of ten-billion and hundred-billion numbers.

Atrium launched in 2017 for startups: fundraising, hiring, M&A paperwork, collaboration software, plus in-house lawyers. The pitch sounded clean: full-stack firm plus tech efficiency, faster and cheaper than traditional counsel. General Catalyst moved fast with a roughly ten-million term sheet; Andreessen later led a ~$65 million Series B; total funding rolled into the seventy-five-million range.

On The Quest side he later said the team hired and chased customers before the product was differentiated. His line was blunt—you cannot skip a company’s R&D phase; try, and you miss being forced to build something different. Very hard to solve that with money.

Fundraising here was not just fuel. It was a throttle: before you knew whether you served lawyers or founders, you could already staff to a hundred or a hundred eighty.

They did not lack time for product. Fundraising gave them the illusion they already should be scaling.

The real lesson is not “fundraising is bad.”

It is this: fundraising lets a company still searching for product inherit the cost structure of a mature one.

The services company ate the software dream

Atrium wore legal-tech clothing. The skeleton was services.

Software sells scale. A law firm sells people’s time.

Atrium bound the two together: software needs ongoing R&D; legal work needs lawyers delivering without end. The first wants marginal cost to fall; the second adds headcount as volume grows.

So the more “successful” the company looked, the sharper one question became: is growth amplifying software—or amplifying labor?

If the answer is labor, the exponential curve VCs love is hard to draw.

Kan told TechCrunch many full-stack vertical firm models would not survive; Atrium never made a real dent in operational efficiency. In January 2020 it cut in-house lawyers to chase purer software and a “professional services network”; clients felt chaos, trust shook; two months later the startup wound down. A standalone firm could continue; the software venture ended.

A year later he tweeted “don’t build a services company”—more work managing everyone; the reward is not there at the end. That is the experiential verdict. What readers can carry farther is the structural clash: when services fight the VC software growth curve, fundraising only amplifies labor cost faster.

And: only work on things where you have intrinsic motivation. He later said more plainly—he barely cared about legal services or legal tech; he was strong at sales and could roll the boulder uphill, but it was still uphill. Money cannot fix “you do not actually want to do this every day.”

Closing early was not surrender. It was the first time he stopped proving himself

Layoffs and shutdown, he later said, were among the hardest business calls he ever made.

The hard part was not noticing Atrium was broken.

It was that he still had money.

Capital sat in the account. Investors were still there. The team was still there. The press would not sentence you to death for another six months. The easy call was to keep hiring, keep shipping, keep telling everyone “give us a little more time.”

Looking back, Kan asked himself: was there still a path—or was he only dragging everyone into the next fire drill?

He chose stop.

The company did not burn the rest to zero. At shutdown, some capital went back to investors.

That was not a finance trick.

It was the first time he accepted that after failure, what most needs protecting may not be the company—it is judgment.

Because once a company can no longer prove why it should keep existing, the only thing tens of millions left can buy is more time.

And more time does not always bring an answer.

Sometimes it only makes you less willing to stop.

After Atrium, he stopped believing “get big” is itself the answer

That is why, when Justin Kan talks about Atrium, what he keeps returning to is not “we should have raised less.”

If the lesson were only raise less, it would be too cheap.

What he really changed was the order of building.

The earlier Justin Kan had already proven he could build something large: Twitch sold to Amazon for about $970 million; he had YC, investors, and startup-world credit; he knew how to hire, raise, and tell a story big enough.

Atrium was the first time he saw:

Those abilities can all be present—and the company can still not be worth continuing.

Because you may not love the problem.

You may not have found a real product.

You may have wrapped services as software without solving the clash between services and scale.

Or you may simply assume—because the last company worked—that the next one must also be big enough.

That is what Atrium left him.

Not “burn less.”

But: do not mistake the ability to get big for proof that this thing deserves to get big.

What he swapped for the second start was the starting stance

So when you look at Justin Kan’s choices later, the questions he cares about have already shifted.

Not: how much can this company raise?
But: why must I do this?

Not: how do we staff to a hundred fast?
But: can the product stand on its own?

Not: can we hold one more round?
But: without the next raise, should this company still exist?

Those three map to Atrium’s three lessons: motive, product, business model.

Twitch taught him what success feels like.

Atrium taught him that the capacity for success is not the same as the right direction.

What seventy-five million really bought was an expensive “no”

The part of Atrium worth writing is not “seventy-five million went up in smoke.”

It is a person who had already won once—who had capital and reputation—and still chose to say:

This is not worth continuing.

For many founders the hardest sentence is not “I will start.”

It is: I have already spent this much, hired this many, told this big a story—why admit it is wrong now?

Justin Kan said it.

He did not try to copy Twitch’s success onto Atrium, and he did not keep proving after failure that he could still build something bigger.

He stopped.

Only then did he start asking again what he actually wanted to do.

That may be Atrium’s real residue:

Building a company is not forcing it large enough. It is knowing, while there is still time, what deserves to continue.