Marc Andreessen's 2026 Outlook: AI Timelines, US vs. China, and The Price of AI - YouTube
Marc Andreessen argues we're in the biggest tech revolution since electricity, but the real story is the "trillion dollar questions" still unanswered—and why venture's advantage is betting on contradictory strategies while Chinese models catch up in 12 months and Americans adopt AI faster than they'll admit.
Read OriginalMy Notes (6)
SaaS is a $600B market, but labor is $40T+ so AI's real TAM is labor not just software
"VC success depends on fundamental technology shifts."
- During major tech transitions, startups can aggressively enter and dominate new categories before incumbents adapt
- Without such shifts, big companies simply outcompete startups, making VC investments difficult to succeed
- The best VC firms are those skilled at identifying and navigating between these technological waves
Core principle in pricing - you don't wanna price by cost, you wanna price by value. Naive thinking is low prices are better for the customer, when higher prices are better because then the vendor can make the product better faster. Most people that buy things aren't looking for the cheapest price they're looking for things that they know are gonna work really well
"The whole theory of venture that we've had from the beginning is that-you know, many people before us have had as well, that's very correct I think-is the whole theory is like the money in venture is made when there's like a fundamental architecture shift, like when there's a fundamental change in the technology landscape. And that's been true for venture basically forever. And the reason is because if you have a fundamental change in technology then you have this period of creativity in which you can have basically aggressive, you know, very aggressive kind of people start these new companies and they have this kind of shot to come in and win categories before big companies can respond. If there's no fundamental change in technology, it's very hard to make startups work because the big companies just end up doing everything. And so venture kind of lives or dies on the basis of these waves, of these transitions. And so there's always this question-I mean, I would just say the best venture capital firms in history, I think, are the ones that were the most aggressive at being able to navigate from wave to wave."
"I think in this business, like of all businesses, you just need to get onto the new thing. It was—I mean quite honestly it was—pretty amazing that most of the venture ecosystem just decided to sit crypto out. And the number of VCs that we talked to between, call it, the release of the Bitcoin white paper in 2009 to the beginning of the crypto war in 2021 who just basically said, "Oh, we're not going to do crypto." I never quite know what to do with the VC who says, "Oh, there's a new wave of technology and I'm very deliberately not going to participate in it." And I'm always like, "Is that not the job?""
lol
"Classic social science question, which is like okay, if you want to understand basically patterns of people, there's basically two ways to understand what people are doing and thinking. One is to ask them and the other is to watch them. And every social scientist, every sociologist will tell you this, which basically is you can ask people—and the way you do that is surveys, focus groups, polls, what they think. But then you can watch them and you can do what's called revealed preferences. Just observe behavior. You can actually watch their behavior and what you often see in many areas of human activity, including politics and many different aspects of society and culture over time, is the answers that you get when you ask people are very different than the answers that you get when you watch them. And the reason is—I mean you could have a bunch of theories as to why this is. The Marxists claim that people have false consciousness. The explanation I believe is just people have opinions on all kinds of things, particularly when they're in a context where they get to express themselves, and they'll have a tendency to express themselves in very heated ways. And then if you just watch their behavior, they're often a lot calmer and a lot more measured and a lot more rational in what they do. And so that's playing out in AI right now, which is if you run a survey or a poll of what, for example, American voters think about AI, they're all in a total panic. It's like, "Oh my god, this is terrible, this is awful, it's going to kill all the jobs, it's going to ruin everything." But if you watch the revealed preferences, they're all using AI."
Summary used for search
TLDR
• The "revealed preferences" gap: Americans poll as terrified of AI but are adopting it faster than any technology in history—using ChatGPT to draft breakup texts, diagnose skin conditions, and save their jobs
• Chinese companies like DeepSeek caught up to GPT-4 level in under 12 months with far fewer resources, suggesting technological leads aren't durable and open source is accelerating knowledge proliferation
• AI application companies are backward-integrating to build their own models and experimenting with value-based pricing ($200-300/month tiers) rather than just tokens-by-the-drink
• The real threat: 1,200+ state-level AI bills could kill American development through downstream liability for open source developers—but federal government is waking up to this as a China competitiveness issue
• Per-unit AI costs are collapsing faster than Moore's Law due to chip competition, model efficiency gains, and hyperscaler buildout—the shortage will become a glut
In Detail
Andreessen frames AI as the culmination of an 80-year alternate path in computing—the neural network approach that was theorized in 1943 but only crystallized with ChatGPT three years ago. The key insight is that we're still in the "trillion dollar questions" phase: open vs closed source, big vs small models, incumbents vs startups. Unlike companies that must commit to one strategy, venture capital's structural advantage is betting on all plausible approaches simultaneously. He's seeing unprecedented revenue growth in AI companies—faster than anything in his career—but believes current product forms are primitive compared to what's coming.
The Chinese catch-up phenomenon is reshaping assumptions about defensibility. DeepSeek (from a hedge fund, not a national champion) reached GPT-4 capability in under 12 months and released it as open source. Now four Chinese companies have effectively caught up, suggesting that once someone proves something is possible, replication happens fast even with fewer resources. This is killing the narrative of permanent leads and forcing DC to recognize AI regulation as a national competitiveness issue rather than just a safety concern.
On business models, the infrastructure layer is playing out better than expected—big tech companies are essentially commoditizing their magic AI through cloud services at usage-based pricing, which is fantastic for startups. But the best AI application companies aren't just "GPT wrappers"—they're backward-integrating to build their own models, using dozens of different models for different tasks, and experimenting with value-based pricing tied to productivity gains rather than just cost-plus. The per-unit economics are collapsing faster than Moore's Law due to chip competition (Nvidia's profits are a "bat signal" drawing competitors), model efficiency gains, and hyperscaler buildout. Meanwhile, the real policy threat is 1,200+ state bills that could assign downstream liability to open source developers—California's SB1047 would have made an academic liable for a nuclear meltdown if their model was used in the plant five years later.