TechFlow Logo
Login/ Sign up
ETH Gas
Gwei
Fear
gas
Interview with Folius Founder Jason Kam: Crypto VCs Will Disappear — Stop Looking for New 100x Stories Next Round, Just Buy the Winners

Interview with Folius Founder Jason Kam: Crypto VCs Will Disappear — Stop Looking for New 100x Stories Next Round, Just Buy the Winners

2026.10.22
Share
Future winners may no longer come from "searching for the next unheard-of new coin," but rather from finding true Winners capable of securing long-term market share in already validated massive markets.

Author: 168X (Host Mr. Z) × Jason Kam (Founder of Folius Ventures)

Crypto has not disappeared, but has moved from the Wild West into the Fintech era: The true massive opportunities are converging onto "everything on-chain" and "dollars entering everything"

On October 21, 2026, 168X invited Jason Kam, Founder of Folius Ventures, for a conversation spanning Crypto VC, Bitcoin, AI, semiconductors, and Chinese tech stocks.

Jason's shift itself is the most noteworthy signal of this market cycle. Since September 2024, Folius has largely halted new primary Crypto investments. Meanwhile, since January/February this year, Jason has shifted approximately 80%–90% of his focus to AI and secondary markets. At the same time, he is more willing than ever over the past year to gradually buy dips in Crypto, especially BTC.

The reason isn't that "Crypto has a new Narrative"; quite the opposite. Jason's view is that the stories that could be told in past cycles have basically been exhausted, and tracks achieving true PMF are rapidly converging. Future winners may no longer come from "finding the next unknown coin," but rather identifying true Winners in already validated, massive markets that can hold long-term market share.

I. Folius Hasn't Invested in Primary Crypto for Two Years: This Cycle Has Shifted from "Sowing" to Harvest

Mr. Z: Folius was founded in a completely different Crypto environment during the last cycle. Looking at where things stand today, what has changed the most in your investment framework?

Jason: Folius can essentially be split into two entities now. One is the legacy fund, and the other functions more like my own holding company and small family office, with some external capital, allowing us to look at stocks, Crypto, or anything.

Jason: Since January/February this year, I've focused roughly 80%–90% of my energy on AI, particularly secondary markets. Over in Crypto, it's mostly been managing existing positions.

Jason: Furthermore, we have basically stopped making primary Crypto investments since September 2024. One reason is that the lifecycle of the old Token Factory model is over. A few people get together, raise funds, do a TGE, issue low float/high FDV tokens, wait for unlocks—the whole playbook is losing its logic.

Jason: Another reason is that our fund structure allows quarterly redemptions for LPs. Such a structure is inherently unsuitable for early-stage projects that take 5 to 8 years to exit.

Jason: So it's more of a harvesting phase now. For past investments, the focus is on realizing returns and managing liquidity, rather than forcing our way into new private deals just to prove we're still VCs.

II. No More Crypto VC by 2030: Only Fintech VC, Liquid Funds, and Multi-Sector Funds Will Remain

Mr. Z: If you look ahead to 2030, do you think the "Crypto VC" category will still exist?

Jason: I definitely see an end to traditional Crypto VCs.

Jason: Liquid funds and prop funds will certainly persist, because Crypto has its own market structure, on-chain data, tokenomics analysis, and historical experience—all of which create a distinct edge. But if you mean traditional-style VCs exclusively focused on primary Crypto investments, I find that increasingly difficult.

Jason: The reason is simple. As Crypto becomes more institutionalized, your real competition shifts to firms like Sequoia, Founders Fund, and USV. Why would a top founder take money from a boutique Crypto fund?

Jason: In the future, Crypto funds will either pivot toward liquid markets and OTC trading, operating more like public market investors; morph into Fintech VCs investing in stablecoins, payments, and neobanks; or completely broaden their mandate to invest in AI, robotics, and biotech.

Jason: Firms like Paradigm and Framework are already moving in this direction. If an investor's only skill used to be "I understand Crypto," that won't be enough going forward.

III. Crypto Ultimately Boils Down to Two Super Themes: Everything On-Chain, and Dollars Entering Everything

Mr. Z: If the stories around infrastructure, Layer 2s, and new L1 chains are gradually fading, what's left for Crypto?

Jason: I think they've been sufficiently defined already; essentially, there are just two things.

Jason: First, everything going on-chain, and providing a more equitable, permissionless marketplace for all financial products. Ultimately, this translates to everything exchanges, trading, speculation, and various assets entering on-chain markets.

Jason: Second, the dollar as a store of value penetrating every corner of global life. What ultimately emerges from this are stablecoins, payments, payment processors, and neobanks.

Jason: We used to talk about sharding, zero-knowledge proofs, and various application layers, but most of them ultimately died. A core reason is that users simply don't care about the underlying technology. Many solutions were just a hammer looking for a nail.

Jason: But trading and money are different. Global financial assets represent a massive TAM, and cross-border/global payments is another enormous market. Crypto's penetration rate in these two sectors remains very low.

Jason: So it's not that Crypto has lost its opportunity; rather, opportunities have converged from 100 vague tracks into two massive themes.

IV. Stop Chasing the Next 100x: True Alpha Comes from Buying Winners at a Discount

Mr. Z: So as an investor, with the tracks so clear now, how should you allocate capital?

Jason: My thinking has grown simpler: just buy the winners.

Jason: In areas like stablecoins, exchanges, and payments, it's becoming increasingly clear who the leaders are. You can research new startups, but why assume a brand-new team will outperform incumbents that already have distribution, liquidity, engineering capacity, and users?

Jason: What truly matters is evaluating whether a winner's market cap relative to adoption is cheap or expensive.

Jason: For example, with Robinhood, look at market cap vs. assets under custody; with Hyperliquid, look at valuation relative to TVL and volume; similar payment companies should be assessed on actual transaction volume processed and user penetration.

Jason: The market caps of these companies fluctuate cyclically around adoption metrics. During bubbles, they seem ridiculously expensive; at troughs, the market acts as if growth will never return.

Jason: Real alpha might not come from discovering the winner, but having the courage to buy when the market temporarily dislikes it.

V. Don't Fight the Incumbent Head-On: The Greatest Lesson from HIP-3 Is That Capital Has Inertia

Mr. Z: If investing is about buying the winner straight up, do founders still have a chance?

Jason: Yes, but don't build an identical copy of the number one player.

Mr. Z: Take HIP-3, for example. Initially, I thought the design was clever, allowing different teams to launch their own markets on Hyperliquid. But in the end, you saw TradeXYZ capture the vast majority of trading volume. It wasn't just about the mechanism being good; it came down to founder relationships, early ecosystem positioning, and liquidity.

Mr. Z: Capital itself carries inertia and stickiness. Once a positive feedback loop forms, it's extremely difficult for a second-place contender to win solely by saying, "I'm building the same product."

Jason: So founders must find a completely different edge. If others do traditional trading, can you do AI-powered intent-driven trading? If others do pure trading, can you make social the new trading interface?

Jason: Finding new entry points within validated large markets is far more important than reinventing a completely novel track nobody needs.

VI. BTC and Crypto Have Diverged: One Is Digital Gold, the Other Is Fintech

Mr. Z: If there's another bull market in this cycle, what narratives drive Bitcoin versus the rest of Crypto?

Jason: I believe we must separate BTC completely from everything else.

Jason: BTC tells the story of digital gold, while the rest of Crypto tells the story of fintech.

Jason: Historically, Bitcoin acted like the sun, with the entire industry orbiting around it. But now look at assets like Hyperliquid—it's correlated with BTC much less than before, instead increasingly mirroring US equities and fintech adoption trends.

Jason: BTC itself is also increasingly decoupling from the historic four-year halving cycle. Annual new supply is minuscule, and compared to ETF inflows and global USD liquidity, the halving's impact on marginal supply and demand continues to weaken.

Jason: Going forward, BTC will resemble gold, slowly appreciating alongside USD cycles and global portfolio allocation. Other tokens, if they generate real cash flow, feature genuine value accrual, have credible management, and show growth, effectively behave more like stocks that trade 24/7.

Jason: So stop using the same framework to value both BTC and HYPE. They are fundamentally two completely different asset classes.

VII. How to Play BTC Over the Next 12–18 Months: Buy the Dips, But Remember the Top 3 Scenarios

Mr. Z: So right now, are you bullish or cautious on BTC?

Jason: Since June this year, my baseline view has been to dollar-cost average into dips over the next 12–18 months. If the market drops, I buy gradually; if it rises, I hold; if it runs too high, I might sell some calls.

Jason: Because after cycling through several markets, everyone knows roughly what a cycle bottom looks like. Long-to-medium term holders at these levels inherently lack strong reasons to sell unless forced.

Jason: I see three potential scenarios that could trigger a major drawdown in BTC.

Jason: The first is extremely strong AI productivity gains. If AI truly drives near-infinite productivity growth, why hold an asset that generates no cash flow? Capital could flock entirely to productivity assets.

Jason: The second is Q-Day. If quantum computing suddenly achieves a breakthrough, it would generate massive FUD regarding the security of current cryptography. Eventually, technical workarounds or forks might emerge, but the market will absolutely price in panic first.

Jason: The third is a true financial crisis and forced liquidation cascade. If the AI cycle crashes and triggers a liquidity crunch, BTC will inevitably fall alongside risk assets.

Jason: Outside of those three scenarios, my thesis remains unchanged: if it drops, buy the dip gradually.

VIII. The Largest Source of Liquidity for the Next Crypto Cycle May Be AI: Employee Exits Alone Could Inject $10–60 Billion

Mr. Z: Why do you say future Crypto cycles will be increasingly dictated by AI cycles?

Jason: Because the wealth effect generated by AI is simply too massive.

Jason: You can run a rough back-of-the-envelope calculation: sum up the potential IPOs and exits of all Silicon Valley AI companies over the next two to three years, then assume these employees eventually allocate 2%–15% of that wealth into Crypto.

Jason: My estimate puts that potential inflow between $10 and $60 billion.

Jason: This figure is even comparable to the cumulative net inflows into BTC ETFs since their inception.

Jason: And this only accounts for employee exits, excluding the broader wealth effect created across the market by rising AI valuations.

Jason: So while AI is exceptionally strong now, it doesn't mean Crypto instantly becomes the best play, since capital is currently chasing AI; but once AI-generated wealth starts realizing, a portion will eventually flow into Crypto.

Jason: The reverse holds true as well. If AI suddenly crashes, risk assets cannot avoid damage in the initial stage.

Jason: The biggest macro variable for the next Crypto cycle may actually lie outside of Crypto altogether.

IX. The Hardest Inflection Point in AI Hardware: Focus Shouldn't Be on Token Prices, but on Revenue Per GPU

Mr. Z: You've spent significant time researching AI and semiconductors. What do you find most difficult to gauge in the AI cycle?

Jason: Early in the year was straightforward. Markets hadn't priced in coding agents yet, so you could buy baskets. Later, when GPU demand wasn't fully understood, it was easier to position. But the questions that genuinely need answering now have become much harder.

Jason: The core question isn't whether "token prices are falling," but whether revenue per GPU is actually declining.

Jason: Token prices have indeed been dropping, but the cost for a single GPU to generate those tokens is also plummeting rapidly. Model efficiency, software optimization, and chip utilization rates are all improving.

Jason: Therefore, falling token prices do not necessarily equate to deteriorating GPU economics.

Jason: Another variable is balancing increased demand from open-source models plus post-training against decreased demand from slowing frontier lab growth—which force wins?

Jason: The real danger lies in waiting until you clearly see GPU rental yields trending downward, by which time smart money may have already traded 1–2 quarters ahead.

Jason: Thus, topping out in AI won't be as simple as watching a moving average. The true inflection points are hidden in revenue per GPU, channel checks, and hyperscaler earnings.

X. The Next Leg of AI May Shift from Chips to Hyperscalers: Odds Are Actually Clearer for MSFT, AMZN, and GOOG

Mr. Z: If semiconductors are getting harder to play, what in AI do you feel comfortable buying now?

Jason: Right now, I actually feel more comfortable buying hyperscalers.

Mr. Z: Because open-source models are getting incredibly capable, enterprises don't need to route every task through the most expensive frontier models. CSPs can integrate various open-source models into their clouds and offer enterprise deployments.

Jason: In that scenario, the real beneficiaries are cloud distributors like Microsoft, Amazon, and Google.

Jason: So if I must play AI right now, I'd lean toward MSFT, AMZN, GOOG, or even certain cloud ETFs. Semis are trickier, as expectations are fully priced in, and I still can't determine which way revenue per GPU will settle.

Jason: This isn't a "set it and forget it for a decade" trade. The pace of open-source progress, the closed-source frontier lead, token pricing, and GPU economics are all dynamic.

Jason: But at least in this current phase, rotating from hardware to cloud service providers offers a clearer logic than blindly buying every chip stock.

XI. China AI Wagers on Just Three: DeepSeek, Tencent, and ByteDance

Mr. Z: Looking at China's AI landscape over the coming years, who are you most inclined to back?

Jason: Right now, I'm leaning toward backing three: DeepSeek, Tencent, and ByteDance.

Jason: What attracts me most to DeepSeek is its crystal-clear strategy: build the most open models with the lowest costs and maximum performance, driving the cost-to-performance ratio to the extreme.

Jason: Tencent doesn't necessarily need to build the world's strongest frontier model. Its real assets are WeChat, gaming, and advertising distribution. AI can significantly boost ad efficiency and drastically slash game development costs, while monetization channels remain firmly under Tencent's control.

Jason: ByteDance plays a different game. It boasts formidable consumer distribution and is among the most aggressive Chinese companies investing in compute power.

Jason: Therefore, China doesn't need to completely surpass the US on frontier models before generating massive AI equity value.

Jason: Models provide capability; distribution is the machine that turns capability into revenue.

XII. ByteDance's Biggest Hidden Alpha: Not Doubao, But Video Generation + App Factory

Mr. Z: Why do you find ByteDance particularly interesting right now?

Jason: The most undervalued aspect of ByteDance, in my view, isn't simply Doubao.

Jason: First is video generation. If its generative video models continue iterating, the gap between AI video and traditional film production will shrink considerably over the next one to two years.

Jason: If a piece of content once required actors, cameras, lighting, post-production, and crews, but future productions can be made by just a handful of people backed by massive compute, the entire entertainment cost curve will be completely reshaped.

Jason: The second, arguably more important factor, is that ByteDance itself operates as an app factory.

Jason: It excels at product development, growth, and distribution, having already proven via TikTok that it can rapidly scale products globally.

Jason: As AI makes app development cheaper and easier, companies possessing the true ability to continuously produce apps and push them to hundreds of millions of users will only grow more valuable.

Jason: So ByteDance's moat isn't just "having a model," but rather the combination of model + data + distribution + growth engine.

XIII. Stop Forcing the AI + Crypto Narrative: Genuine AI Researchers Simply Don't Care About Crypto

Mr. Z: So where does true AI + Crypto intersection actually lie?

Jason: Honestly, nobody knows.

Jason: Many serious AI researchers simply don't care about Crypto. Conversely, crypto natives are the ones constantly trying to shoehorn AI and tokens together.

Jason: That doesn't mean zero opportunity. Areas like privacy, private inference, and ventures like Venice address genuine needs; payments, stablecoins, and financial markets may also naturally see more machine-to-machine transactions driven by agents.

Jason: But I won't assume a project has inherent value just because it slaps "AI + Crypto" on its pitch deck.

Jason: The core problem with AI + Crypto has always been that the narrative is too easy to construct, while finding true PMF is brutally difficult.

Jason: Crypto deserves a place only if it genuinely improves AI products, not just to force-feed AI into a token issuance scheme.

XIV. If the US Enters a K-Shaped Downturn: PDD + Agents May Actually Serve as Defensive Assets

Mr. Z: Valuations for Chinese internet stocks are already rock bottom. If the US or globe truly enters a K-shaped economic downturn, where is the next opportunity?

Jason: I evaluate A-shares, Hong Kong listings, and Chinese ADS separately, as each faces distinct liquidity dynamics. However, certain individual companies are dirt-cheap.

Jason: Take PDD—I hold a personal position. I like it not just for the valuation, but because it may possess inherent advantages precisely during a K-shaped downturn.

Jason: As consumers increasingly chase lower prices, compounded by AI agents scouring the web for the cheapest, most efficient products, platforms like PDD that have already optimized low-cost supply chains to the extreme could actually get strengthened.

Jason: Look back at the 2007–2009 US financial crisis; extreme-discount retailers didn't collapse the way high-end consumer brands did.

Jason: So if AI ultimately does trigger divergence in employment, income, and consumption patterns, low-cost distribution paired with agents might represent a fascinating form of downside protection.

XV. Three Things Worth Studying Over the Next 12 Months: Cheap Intelligence, Post-AI Cycle Plays, and Robotics

Mr. Z: Finally, if you could give investors only three directions worth researching over the next 12 months, what would they be?

Jason: First, I'm keen to figure out who will ultimately convert this cost decline of increasingly cheap intelligence into their own profits.

Jason: Tokens are getting cheaper, AI capabilities are commoditizing, and inevitably someone will purchase this inexpensive intelligence and transform it into massive enterprise value. It could be PE firms, software companies, or entirely novel business models.

Jason: Second, if AI moderation sets in mid-cycle due to declining revenue per GPU, what exactly should be bought at that point? Everyone knows how to long AI right now, but few have seriously considered who wins the post-AI cycle.

Jason: Third, I hold a strong conviction that AI will unlock robotics and the physical world.

Jason: The next market cycle will undoubtedly start pricing robotics more seriously. The question isn't whether robotics becomes the narrative, but which company can successfully convert AI intelligence into actual physical productivity.

Jason: So over the next 12 months, the three things I genuinely want to study are: who captures cheap intelligence, who wins the post-AI cycle, and who dominates robotics.

Mr. Z: To close this episode, I condense Jason's framework into five key takeaways. First, traditional Crypto VC is losing its necessity; what remains will be Fintech VCs, liquid funds, or investment platforms spanning AI, robotics, and public markets. Second, Crypto's mega-opportunities have consolidated into two themes: everything on-chain and dollars permeating daily life. Instead of chasing a hundred new narratives, it's better to buy proven winners directly at cycle troughs. Third, BTC is gradually decoupling from the broader Crypto fintech ecosystem to become an independent digital gold trade; for the next 12–18 months, the halving matters less than AI's wealth effect, Q-Day risks, and liquidity crises. Fourth, AI's greatest impact on Crypto may not be AI agent token launches, but rather how much wealth AI creates and how much ultimately flows into Crypto. Fifth, true alpha in the next phase is already shifting from "will AI keep pumping?" to "who will monetize cheap intelligence, who wins the post-AI cycle, and which robotics company prevails." Thank you, Jason Kam, for joining 168X!