PRODUCT HUNT — LAUNCHING SOON 🙌

The 1752vc Pitch Deck Analyzer goes live on Product Hunt in two weeks. Get on the launch list and we'll send you the link the moment we're live — plus 5 free deck reviews if you join the conversation on the day.

👉 Count me in

The most dangerous thing you can hand a pre-seed founder is $50 million.

Not because they'll waste it — though they will. Because it removes the one thing that reliably produces breakthrough thinking: a back against the wall.

Talk to enough founders who actually made it and you'll hear the same story on repeat. "We had one week of payroll left. The company was dead. And then—"

And then they innovated. Not because they got smarter overnight. Because they had no other option.

FedEx exists because Fred Smith took the company's last $5,000 to a Vegas blackjack table to cover a fuel bill. Airbnb exists because the founders were so broke they sold novelty cereal boxes to stay alive — and in the process learned the scrappiness that defined the company's culture for a decade. Instagram was a failing check-in app called Burbn that stripped itself down to a single feature when the runway math stopped working. These aren't cute origin stories. They're the mechanism.

Constraint isn't the obstacle to innovation. It's the input.

Here's what's happening right now: 27 seed rounds of $100 million or more have been announced globally since the start of 2025. Seed rounds. Pre-product, in many cases pre-anything — priced on pedigree before there's a single thing to measure.

An ex-OpenAI badge and a deck now buys you a nine-figure round and a unicorn valuation on day zero.

And it's not just the frontier labs anymore. The mega-seed has trickled down. Application-layer companies with no revenue are raising $30M, $50M, $80M "seeds" because a big fund needs to deploy and a hot resume walked in the door. The round size stopped being a function of what the company needs and became a function of what the fund needs. That's the tell. When check size is set by the investor's deployment schedule instead of the founder's milestones, nobody in the room is pricing risk anymore.

Let me be clear about a couple things: I don't blame the founders. If the market is handing out $100M at a $1.5B cap for a team and a thesis — get yours. Take the money. Capital is cheap when it's offered and brutally expensive when it's needed, and every founder who lived through 2022-2023 knows the window can slam shut without warning.

And I don't blame the VCs either. The math on our side has changed. AI has blown the ceiling off potential TAM — software is no longer selling tools, it's selling work itself. Fund sizes are up accordingly. And the exit that used to define a generational outcome has been repriced: the prize is no longer a $10B IPO, it's a trillion-dollar company. If you believe a team has even a small shot at an outcome above $1T, a $100M seed check is rational underwriting, not recklessness. The power law got steeper, and everyone is pricing to the new tail.

None of that is the problem.

The problem is what happens next.

When you have 18 months of runway and a problem, you think. When you have 8 years of runway and a problem, you spend.

Churn is up? Hire three more sales reps to outrun it. Product isn't landing? Buy more ads. Infrastructure costs are eating you alive? Who cares — throw compute at it. Can't crack enterprise sales? Acqui-hire a team that supposedly can.

Every one of those is capital solving a problem that should never be solved with capital. The churn was a product problem. The CAC was a positioning problem. The infra bill was an engineering problem. Money didn't fix any of them. It just made them quieter — and more expensive.

Here's the reality: money doesn't speed up learning. It makes bad ideas more expensive.

Think about what a constraint actually does inside a company. It forces prioritization — you cannot do everything, so you're forced to figure out what actually matters. It forces contact with reality — you can't afford to guess, so you talk to customers. It forces creativity — the obvious solution is off the table, so you find the non-obvious one. Strip the constraint away and none of that pressure exists. The company doesn't get smarter. It gets bigger. Those are not the same thing, and confusing them is how you burn $80M learning what a bootstrapped competitor learned for $80K.

The data has said this for over a decade. The Startup Genome project found that premature scaling — spending ahead of validated learning — was the number one killer of high-growth startups, implicated in roughly 70% of failures. Startups that scaled properly grew 20x faster than those that scaled early. And in the last downturn, one analysis of 431 failed VC-backed startups found they raised a combined $17.5 billion before dying. Median raise: $11 million. They didn't die of starvation. They died of abundance.

We ran this experiment at scale already. It was called ZIRP. A decade of free money produced a generation of companies that mistook fundraising for progress and headcount for momentum — 3,000-person startups with no path to profitability, burning $20M a month to buy growth that evaporated the moment the discounts stopped. When rates rose and the tide went out, the companies that survived weren't the best-funded ones. They were the ones that had kept operating like the money might run out.

The overfunded companies don't iterate. They don't confront. They don't learn.

They deploy.

Want the cleanest natural experiment in constraint-driven innovation? Look at China's chip stack.

US export controls cut Chinese labs off from Nvidia's best GPUs. The intent was to freeze them out of frontier AI. The result? DeepSeek — banned from top-tier hardware, running on deliberately handicapped H800s — was forced into compute-efficient training methods and trained a GPT-4-class model for a reported $5.6 million. While US labs were raising tens of billions to brute-force scale, a constrained team rewrote the efficiency frontier because they had no other move available.

And the constraint kept compounding. Because they couldn't buy their way to bigger clusters, they innovated on architecture — mixture-of-experts routing, aggressive distillation, training tricks the flush labs had no reason to bother with. Then they open-sourced the results and collapsed the price of inference for everyone. One sanctioned lab did more to bend the cost curve of AI than a hundred billion dollars of well-capitalized competition. The whole ecosystem is now standing on techniques that exist only because someone couldn't write a check.

The wall didn't stop them. The wall was the R&D roadmap.

Meanwhile, what did unlimited capital buy the incumbents? Bigger clusters running the same playbook. The constrained player innovated on method. The capitalized players innovated on invoice size.

The counterargument is real, so let's give it its due. Some businesses genuinely require massive upfront capital. Frontier models, chips, biotech, defense — you cannot bootstrap a foundry. Capital as a moat is a legitimate strategy when compute or regulatory approval is the actual product. And in a winner-take-most market, speed funded by capital can beat elegance funded by scarcity. Uber's war chest wasn't waste; it was the strategy, and it worked.

But.

That describes maybe 2% of the companies raising like it describes them. The other 98% are application-layer startups raising frontier-lab money to avoid frontier-lab discipline. They're buying the costume of a capital-intensive business without the physics that justify it. If your product is software on top of someone else's model, you are not CoreWeave. You do not need a war chest. You need ten obsessed customers and the discipline to find out why the eleventh said no.

And here's the part nobody says out loud: raising nine figures pre-product surrenders the one advantage a startup actually has. Google can outspend you infinitely. Google cannot out-desperate you. The incumbent's weakness has never been money — it's that nobody inside the building has their livelihood riding on this quarter's product decision. Desperation is the startup's moat. Raise enough money and you've filled in your own moat, then challenged the incumbent to a spending contest across it.

For founders: raise what the market gives you, but run the company like you didn't. Before capital touches any problem, ask one question — is this a problem money solves, or a problem money hides? Hiring ahead of a signed enterprise contract: money solving. Hiring to outrun churn you don't understand: money hiding. Fake constraints if you have to. Give teams fixed budgets and impossible timelines. Milestone-gate your own treasury — put the raise in tranches in your own head even if your investors didn't. The discipline you'd be forced into at $2M raised is the discipline that compounds at $100M raised.

For LPs: when a GP tells you the $200M seed round "de-risks" the company, ask what the Startup Genome data says about spending ahead of learning. Ask why the most efficient AI lab in the world was the one that couldn't buy GPUs. Ask whether that check size was set by the company's milestones or the fund's deployment clock. Deployment is not diligence. Size is not safety.

The bottom line: nobody tells the story of the round that saved them. They tell the story of the week they almost died — because that's the week they became a real company.

Your back against the wall isn't the thing to escape.

It's the thing to keep.