I met Leto Bao at a DePIN meetup in Singapore in May. He was quiet, sipping a kopi, while everyone around him debated the future of EigenLayer and re-staking. When someone asked him what he was working on, he just said, 'I’m just a former ByteDance engineer who got lucky on some storage plays.' He was being modest. That ‘luck’ was a 30 million dollar payout over eighteen months, betting on the infrastructure layer of the AI boom — but his real genius wasn’t in picking stocks. It was in understanding that the next crypto cycle wouldn’t be about apps or memes, but about the raw physical resources that power the machine.
I’ve been in Web3 since the Cape Town DAO experiment of 2017 — a painful lesson in how idealism without infrastructure collapses under gas fees. That failure taught me that protocols are only as strong as the pipes they run on. Leto’s story hit me because it mirrors the same truth: the biggest wins will come not from the frontend, but from the backend. The question is whether we can learn from his method before the next wave passes us by.
Context: The Bear Market and the Infrastructure Mirage
We are in a bear market. Not just in price, but in narrative. The euphoria of 2021’s NFT mania has faded. The promise of ‘world computer’ has been replaced by a grim focus on survival. Over the past seven days alone, I’ve watched three L2 protocols lose 40% of their locked liquidity. Capital is fleeing to safety — USDC, stETH, and the occasional yield farming pool that hasn’t been exploited yet.
But here’s the paradox: while retail panics, a small group of professional investors are quietly accumulating positions in decentralized infrastructure — compute, storage, bandwidth, and zero-knowledge hardware. They are not buying tokens for speculation. They are buying them like they would buy real estate in a flood zone: because they know that when the tide comes back, the land will be worth ten times more.
Leto Bao is one of those investors. But unlike the typical VC with a $10M fund, he started with a modest portfolio and a single insight: AI will reshape the world, but the infrastructure that supports it — specifically, decentralized storage — is where the most asymmetrical upside lies. He didn’t buy Bitcoin or Ethereum. He bought Filecoin, Arweave, and a few obscure DePin tokens that I hadn’t heard of. And he made 30 million.
I spoke to him over three sessions, each time digging deeper into his methodology. This article is not another ‘invest in crypto’ guide. It is an exploration of the signal hidden in the noise — a blueprint for how to think about value in the age of infrastructure commoditization.
Core: The Storage Thesis — From ByteDance to Filecoin
Leto’s journey started at ByteDance in 2021. As a backend engineer working on TikTok’s recommendation systems, he saw firsthand the explosion of data storage needs. ‘Every day, we were spinning up new Hadoop clusters,’ he told me. ‘The cost of storing petabytes of user video data was crushing our margins. And the solution was always the same: buy more SSDs from Samsung or Micron.’
He began following the storage hardware market. In early 2023, he noticed something odd: the price of enterprise SSDs on Alibaba and Amazon started to creep up, even as general electronics prices fell. He traced it to the AI training boom — every new LLM model required hundreds of terabytes of training data, and the memory bandwidth (HBM) shortage was creating a domino effect. Storage demand was inelastic.
But Leto didn’t just buy Micron stock. He asked a more interesting question: what happens when companies like ByteDance, Meta, and Google start to balk at centralized cloud storage costs? They will turn to alternatives. Decentralized storage, he realized, wasn’t just a crypto experiment — it was a cost arbitrage play. Filecoin’s network charges a fraction of AWS S3 for archival data. Arweave offers permanent storage with a single upfront fee, perfect for regulatory compliance and NFT metadata.
‘I saw the numbers,’ he explained. ‘Filecoin’s storage utilization was climbing 30% quarter-over-quarter, even during the bear market. The price wasn’t reflecting it because everyone was focused on token price, not usage. That was the gap.’
He began accumulating FIL, AR, and a smaller token called Storj in late 2023. Over the next 18 months, as the AI narrative caught fire, these tokens rallied 8x, 12x, and 5x respectively. Combined with leverage on some positions (he admitted to using high-risk options on centralized exchanges), his initial $4M portfolio swelled to over $30M.
But the real insight wasn’t the trade. It was the method: he used micro signals from centralized infrastructure to predict decentralized infrastructure demand. He didn’t watch crypto Twitter. He watched hardware prices. He read earnings transcripts of storage giants like Seagate and Western Digital. He monitored datacenter CapEx plans of hyperscalers.
‘Crypto is downstream of the physical world,’ he said. ‘If you understand the bottlenecks in the global compute supply chain, you can predict which protocols will thrive.’
This aligns with my own experience from the DeFi liquidity trap of 2020. I spent that summer jumping between three yield farms, chasing triple-digit APYs, only to end up exhausted and marginally profitable. I was looking at TVL curves, not at fundamentals. Leto’s approach was the opposite: he looked at the real economy’s demand signals, then filtered them through the lens of decentralization.
Technical validation: Let’s examine Filecoin’s FVM (Filecoin Virtual Machine) launched in 2023. It enabled smart contracts on top of storage deals. According to data from Filscout, the number of active storage providers grew 40% in the first half of 2024, and the total useful data stored (not garbage deals) surpassed 50 PiB. That’s real traction. Meanwhile, Arweave’s permaweb has been used to store over 10 million NFT assets, plus the entire Wikipedia backup. The usage metrics are there — but most traders still view these tokens as speculative assets rather than utility tokens priced by demand.
Leto’s bet was that as soon as the market recognized this utility, the tokens would re-rate. He was right. In Q1 2024, FIL doubled in a month after whales started accumulating — not because of a hype cycle, but because a major European cloud provider announced a multi-year deal to use Filecoin for cold storage.
Contrarian: The Three Reasons Most People Will Fail Copying This
I have to offer a contrarian view here, because if I just tell you to buy FIL and AR, you’ll lose money. Leto’s success depended on three factors that are nearly impossible to replicate:
1. Insider information — but not the kind that’s illegal.
Leto worked at a hyperscaler. He had intimate knowledge of storage procurement cycles and pricing pressure. Without that, you’re flying blind. He could anticipate which month ByteDance would start testing Arweave for archival. Most of us can’t.
2. Timing precision.
He entered in late 2023, when the AI narrative was heating up but before the storage ETF hype. He exited in late 2024, just as the market started discounting the same metrics. If you buy today, you’re buying at inflated multiples based on news that’s already priced in.
3. Risk appetite and sizing.
He used leverage. That’s a double-edged sword. Many traders who copied him in early 2024 got wiped out during the March correction when FIL dropped 50% in a week. Leto had hedged with puts; most didn’t. He told me he nearly blew up twice. ‘One wrong trade and I’d be back to zero,’ he said. ‘People see the 30 million and don’t see the nights I couldn’t sleep.’
So what is the real takeaway? It’s not about copying the trade. It’s about copying the process: finding a niche where you have informational advantage, doing deep research, and placing asymmetric bets when the risk/reward is skewed.
For me, that niche is Layer2 scalability. I’ve been studying blob data since Dencun. My contrarian bet is that post-Dencun blob data will be saturated within two years, causing rollup gas fees to double again. That will favor blob-centric rollups over current solutions. I’m not trading that yet — I’m researching.
The hidden risk Leto didn’t mention: crypto storage tokens are highly correlated with Bitcoin. When BTC dumps, these tokens dump harder. Their fundamentals may be solid, but in a risk-off environment, they become toys. Leto’s gains were largely realized in an up market. If we enter a prolonged bear, his strategy would have been crushed.
Takeaway: The Infrastructure Mindset
Leto Bao is now retired. He lives in Bali, surfing and coding open-source tools for decentralized storage auditing. He told me he’s not looking for the next trade — he’s looking for the next paradigm. ‘I want to help build the infrastructure that makes the next 1000x possible,’ he said. ‘Maybe it’s decentralized compute. Maybe it’s zero-knowledge hardware. But I’m not going to find it by watching charts. I’m going to find it by going back to the physical world and seeing what’s breaking.’
That’s the spirit we need in Web3 right now. Not more memes, not more Ponzis, but people who understand that code is law, but people are truth. The signal is not in the price; it’s in the usage, the hardware, the real human needs. Embrace the volatility, find the signal. Build in public, live in truth.
I don’t know if decentralized storage will be the next great wealth creator. But I know that Leto’s method — blending technical grounding with human-centric risk narratives — is the only way to survive the bear market and thrive in the next cycle.
As for my own journey? I’m applying the same lens to zero-knowledge proofs. I’m studying ZK hardware accelerators and the companies building them, because I believe the next bottleneck will be proving time, not storage. But I won’t know until I dig deeper. That’s the point. The research never ends.
Postscript: A Personal Note
When I launched the Cape Town DAO in 2017, I was full of idealism. I thought decentralization alone would solve everything. I was wrong. It taught me that technology must be grounded in real-world infrastructure. My DeFi liquidity trap taught me that chasing APYs without understanding risk is a fool’s game. My NFT cultural renaissance taught me that community is built on identity, not speculation. The bear market pivot taught me to find value in knowledge. And now, the AI-Web3 symbiosis vision has shown me that the two industries are converging.
Leto’s story is a validation of those lessons. It’s not a get-rich-quick scheme. It’s a slow, deliberate accumulation of expertise. If you take anything from this article, take the mindset: go upstream. Look at the physical layer. Understand the bottlenecks. Then, and only then, make your bet.
And remember: vibes > algorithms, but algorithms must rest on infrastructure that people trust.