Every weekend, there’s a new “hero” tweet that takes over venture capital group chats.
This past weekend was clearly slow on the news front, because what I initially read as a fairly standard take ended up dominating my timeline.
The tweet came from Martin Casado, GP at Andreessen Horowitz, who runs their $1.25B infra practice:
The idea that non consensus investing is where the alpha is, is actually quite dangerous in the early stage. Follow on capital tends to be more and more consensus aligned.
For whatever reason, this concept triggered a bunch of investors.
We’re living in an era where the “consensus categories” like AI, robotics, crypto, bio, defense, autonomous systems, manufacturing are big enough to carry entire generations of startups. You don’t need to invent a new category to generate massive outcomes.
It wasn’t always like this. For decades, VCs sat around waiting for the one platform shift (PC, internet, mobile). Now? We have five or six happening simultaneously.
When I started angel investing in 2014, venture capital felt predictable.
Mobile was exploding, on-demand apps were raising boatloads of money (usually with terrible unit economics), and cloud was starting to compound. It was the perfect era to learn the ropes and see how quickly categories get crowded.
Martin’s larger point, as I read the tweet and the replies, is that very few categories are truly non-consensus anymore. Our industry has matured.
Some investors are better at sourcing, picking, and winning. That doesn’t make them “non-consensus”. It just makes them great.
Yes, you can still find non-consensus companies inside consensus categories. But at some point, calling ourselves “non-consensus investors” is just glamorizing the job.
Take American Dynamism as an example, a venture category branded just a few years ago by A16Z that now feels like an industry in itself.
Palantir (2003)
Truly contrarian. Nobody else was building forward-deployed CIA software. Palantir was a one-of-one company that probably only Alex Karp, Joe Lonsdale, and Peter Thiel could have created. Two decades later, it’s more relevant than ever with a $370B market cap.
Anduril (2017)
Equally contrarian, but for different reasons. At the height of Trump-hate in Silicon Valley, backing autonomous defense systems was a career risk. Elad Gil has said he actually got furious emails from friends after his Anduril investment became public. Today Anduril is a decacorn. Only eight years ago, it was radioactive.
2025
Fast forward and American Dynamism is a full-blown investment category. Hundreds of startups and dozens of funds are chasing defense, aerospace, and manufacturing. I live 15 minutes from El Segundo which has famously become a dense hub of these companies.
Today’s defense investors aren’t contrarian. It’s one of the most consensus categories you’ll find.
Does it matter? Not really. Some companies will hit. Most won’t. That’s venture capital.
Today, being non-consensus means cutting through the noise, tuning out VC group-chat gossip, and backing the startups that still feel too weird or risky, even if their category already feels crowded.
In other words: being a non-consensus investor inside consensus.
Crypto is the perfect example. It’s a $2.5T market, Bitcoin’s been running for fifteen years, and BTC + Ethereum already settle trillions annually. By any standard, it’s consensus, yet still widely misunderstood.
Plenty of VCs (especially in the Bay Area) still can’t see past the FTX blow-up in 2022 even three years later.
Meanwhile, Stripe is betting its roadmap on stablecoins, even building its own payments-focused L1. And Robinhood, now a $95B company, is basically a crypto company in disguise.
The paradox is that crypto is consensus as a category today to those who pay attention, but the opportunity is to be non-consensus about its timing and trajectory while making bets that most investment committees are too slow or scared to underwrite.
Peter Thiel is piling money into Ethereum right now while most of Silicon Valley is still trauma-bonding over FTX.
Martin’s observation hit a nerve because it points to something structural:
“Follow on capital tends to be more and more consensus aligned.”
Growth investors aren’t less imaginative, they’re more efficient at finding outliers, and they have massive funds to deploy. Sure, that makes venture feel less romantic. When you’re managing tens of billions, romance doesn’t scale.
More importantly, the best companies are known earlier than ever. Scouts spray $25K checks, report back to their multi-stage overlords, and any information edge evaporates within minutes of a founder update.
Sub-verticals become consensus at lightning speed. AI legal, AI healthcare, AI customer service. There are few secrets in Silicon Valley anymore.
VC right now is noisy AF. If you’re running a contrarian playbook and it works, you deserve the win. But most VCs aren’t Peter Thiel.
I came across a thread last week reviewing 61 (!) AI coding agents and sent it around to my teammates at Chapter One.
We already knew the category was crowded, but seeing dozens of venture-funded startups lined up with screenshots made it feel ridiculous.
“I’ve tried all 61 AI coding agents & IDEs: Emergent, CodeRabbit, Cursor, Copilot, Replit, Devin, Github Spark, Figma Make & more.”
The author included this emoji in the headline:
Accurate. That’s exactly how it feels to be a seed investor right now.
After about two minutes of that thread, I wanted to clear my head and jump in the Pacific Ocean for a cold plunge (though at this point, does taking cold plunges make me consensus?).
I can count the number of truly non-consensus investors on one or two hands. Not because the rest aren’t smart (in fact many are really f’ing smart), but because venture is now hyper-networked, hyper-efficient, and hyper-competitive.
Alpha is still out there. But it doesn’t come from LARPing contrarian. It comes from knowing what you believe in and moving faster than others.
It comes from sitting through thousands of meetings that mostly feel like wasted time. From enduring dry spells where you wonder if you’re even good at this.
Then one day, you meet a truly special founder. That’s when it matters to move faster than the funds waiting for their Monday partner meetings. To decide, commit, and win the right to invest before the market catches on.
And it will catch on fast if you don’t.
In the rarest cases, you’ll back founders who invent categories so big they look obvious to Sand Hill Road only in hindsight. But that isn’t usually about being “consensus” or “non-consensus.” It’s a messy mix of hard work, skill, and luck.
In 2025, venture capital is hyper-networked and hyper-efficient. Scouts live-tweet deal flow. Firms pitch LPs on “unique access.” Everyone swears they’re contrarian.
Behind the noise, the truth is it’s harder than ever to be non-consensus. But it’s also the most exciting time to invest.
There were years when we all sat around waiting for the next platform shift. The number of incredible companies being built right now is staggering.
Anyway, I should stop writing and get back to work. Another legal AI deck just hit my inbox.
You actually answered your own question: to make generational/outsized returns, you need to be contrarian (think Open AI first round, Palantir, Uber's seed, bitcoin, ETH ICO etc...).
You can make money with non-contrarian bets, but it is much more difficult because:
- entry valuations tend to go up
- more rounds raised = more dilution along the way
- more competition for the startup = less odds of creating a generational company
- more competition for the round = lower allocations/harder allocations
Thoughtful!
I don't understand this need for the startup ecosystem to be "non consensus". Non consensus means willing to take a bet that no one (or few) would be willing to make. By definition this excludes all already attractive / high-growth / proven (or well on their way to proving) businesses. Why would any want to pass on those?
So while there might be some outliers in being non-consensus, and it's a muscle to maintain, I agree with Martin's original point about consensus — by and large, you have to chase the heat, and win on different abilities than being non-consensus.