Back to Insights
AI & Investing

Claude Opus 4.5's Breakout Moment & Investing in 2026 with Qiao Wang

57 min Empire (Blockworks) Jan 2026

Key Takeaways

  • Claude Opus 4.5 can "one-shot" entire projects — The biggest difference is that before AI could build demos, but the last 5% required human engineers. Now with a clear spec, Opus can complete the entire project.
  • Chatbot and coding assistant are merging — When you talk to a chatbot, it writes code on the fly. When you code, you write plain English. They're becoming the same thing.
  • 3-4x productivity gains for startups — Every Alliance cohort reports higher productivity gains. The latest cohort reported 3-4x improvement. For big tech, moats still exist (data, switching costs, ecosystem).
  • A 1-2 person unicorn is possible in 2026 — Multiple people are running $10M ARR businesses solo. These are often ex-engineers from Meta/Uber who got sick of bureaucracy.
  • Qiao built a "digital Buffett" stock picker — Uses deep research to gather facts, then a reasoning model to emulate how Buffett/Munger would evaluate investments. 4 of 10 outputs matched Berkshire's actual portfolio.
  • Portfolio: 40% cash, 30% Bitcoin, 30% stocks — He's cautious because valuations are near historic highs. Crypto tokens ex-Bitcoin are less than 1% of his portfolio.
  • Adobe might be the "Google trade" of 2026 — Trading at 12x P/E despite enterprise lock-in and pricing power. The market thinks AI will kill it; Qiao disagrees.

The Opus 4.5 Breakthrough

Qiao Wang, co-founder of Alliance DAO and former quant trader, lists his "holy shit, it's over" moments in AI: the release of ChatGPT, the first reasoning model (o1), Tesla FSD V13, and now Claude Opus 4.5. What makes Opus different?

The answer is the last 5%. Before, AI could put together a demo quickly, but the final stretch—bugs, corner cases, polish—always required a skilled human engineer. With Opus 4.5, if your spec is clear and comprehensive, it can one-shot the entire project.

"The line between general chatbot and coding is blurring. When you talk to a chatbot, it writes code on the fly. When you code, you write plain English. They're merging."
— Qiao Wang

The Death of Software Moats

For early-stage startups, there are basically no moats anymore. Software is too easy to copy. But the giants—Google, Apple, Microsoft, Amazon—still have their defenses: proprietary data, developer ecosystems, switching costs, and mission-critical enterprise integration.

This is why Qiao sees Adobe as potentially the Google trade of 2026. The market thinks AI image and video models will kill Adobe. Qiao disagrees: Adobe's moat is enterprise integration, not consumer innovation. Creative professionals have years of muscle memory and their assets stored in Adobe's cloud. Switching cost is enormous.

The AI-Powered Stock Picker

Qiao spent months building a "digital clone" of Warren Buffett, Charlie Munger, Howard Marks, and Stanley Druckenmiller. Here's how it works:

First, deep research gathers facts and data on thousands of tickers using a six-step framework that emulates what these legendary investors would care about. Then, a reasoning model makes the final investment decision. The key insight: deep research is great for gathering facts but can hallucinate. Reasoning models are better at actual decision-making once you feed them verified data.

"I specifically told the prompt: do not use in any circumstance the current or past portfolio information of Berkshire Hathaway. Yet 4 of the 10 stocks it recommended are actually in Berkshire's current portfolio."
— Qiao Wang

The stocks his AI recommends? Google, Chubb (insurance), and Adobe. He runs the same prompt multiple times and averages the outputs—if all five runs say buy, he has strong conviction.

Portfolio Construction for 2026

Qiao's current allocation reveals his cautious stance:

40% cash — Valuations are near historic highs. He'd rather sleep well than maximize returns.

~30% Bitcoin — He still holds a lot of BTC but would face massive tax consequences if he sold.

~30% stocks — Google (biggest position), Tencent, Amazon, Adobe, and some Eli Lilly.

<1% crypto tokens — Crypto ex-Bitcoin isn't attractive yet. He's waiting for a 2022-style opportunity.

His thesis on specific stocks: Google has the shopping data moat that ChatGPT can't kill. Amazon has flat human headcount over five years while robot headcount grows 20-30% annually. Tencent is "a boring old Chinese company" but incredibly high quality.

The One-Person Unicorn

The most striking prediction: a 1-2 person billion-dollar company is possible in 2026. It's probably already started—just not at that valuation yet. Qiao knows multiple people running $10M ARR businesses completely solo.

These founders aren't talking about what they're doing. For the first time, successful startups don't want to shout to the world—because software is too easy to copy now. The moat isn't the code; it's staying quiet.

"Everyone can and should write code. And by writing code, I mean specifically automating some part of their own life or work. If they don't, they're going to be left behind."
— Qiao Wang

What AI Can't Replace

Despite his AI enthusiasm, Qiao doesn't use AI for writing. Not because it can't write—but because writing is thinking. The process of translating thoughts into words over multiple weeks clarifies his thinking in a way that AI output simply can't.

And on social media, authenticity stands out. His tweets are lowercase, full of typos—because if you're on Twitter these days, everything AI-generated is slop. Human writing feels more real, and that's more valuable than polish.

What if you had invested in Google?

Qiao's biggest trade of 2025. See what your returns could be.

Calculate GOOGL Returns