We didn’t just witness a change in venture capital strategy; we watched the underlying social contract of the industry get rewritten in real-time. This isn’t an investment thesis on Sequoia. It’s an anthropological observation of what happens when the most successful institution in the old economy decides it must become the most aggressive predator in the new one.
When Roelof Botha stepped back, and Doug Leone followed, the ascension of Alfred Lin and Jess Lee wasn't just a changing of the guard. It was a signal that the firm known for betting on the "garage" founder was now making a calculated, existential bet on the machine itself. We didn’t just see Sequoia invest in AI; we saw them capitulate to its logic — a logic that demands scale, speed, and capital at a velocity that makes even their legendary growth funds look like they're moving in slow motion.
This shift isn't solely about coding or chips. It's about the codification of trust. For decades, venture capital operated on a simple heuristic: founders are the risk, and the market is the reward. Sequoia perfected the art of identifying the unpolished human diamond. Now, the country's most influential capital allocators are looking at a different kind of diamond — one that is synthetic, probabilistic, and exponentially scalable. The question is whether the volatility of this new asset class fits the stewardship model they built.
We've moved from the era of the "founder-led" company to the era of the "compute-constrained" company. The most profound shift is how Sequoia, under Lin and Grady — a founder and a data scientist — is treating capital as a raw material, not just a resource. In my years auditing smart contracts, I learned that you can trace the intent of a developer by looking at how they handle edge cases. You can do the same with a venture firm by looking at how they handle uncertainty. The recent push for aggressive AI investments is Sequoia handling the edge case of our generation: the possibility that they might miss the next paradigm shift.
The Open-Source Doctrine vs. The Closed-Loop Machine
To understand why this is a "blockchain" story, you have to understand the foundational tension between the two technologies. My journey through the Ethereum core dev trenches taught me that one of the priceless primitives of that ecosystem is open-source composability. You audit a contract, you fork it, you understand it. You stare at the code, you see the vulnerabilities, but you also see the honesty of the system. It’s a transparent ledger of logic.
AI, as funded by the latest wave of capital, is fundamentally opaque. The power of a model like GPT-5 or whatever comes next isn't in its architecture; it's in its weights — trillions of numbers that encode a statistical shadow of human thought. It is the opposite of a public good. It is a proprietary fortress. And when Sequoia invests billions into infrastructure companies, they are erecting the walls of that fortress.
From core dev trenches to community heartbeat, I've always believed that true innovation democratizes the means of production. Bitcoin gave us a monetary network we don't need to ask permission to use. AI, as it's being financed right now, is dangerously close to becoming a utility you have to pay rent for, controlled by a central authority. The AI gods of the compute layer are becoming the new sovereigns.
Sequoia is not merely betting on a few apps. They are betting big on the picks-and-shovels: Nvidia chips, cloud infrastructure, energy grids. They are investing in the "picks and shovels" of a new gold rush. This means they are investing in centralization. If AI creates the most dynamic economic value since the internet, whoever controls the largest concentration of compute — and the capital to buy it — controls the entire value chain. That is the macro thesis.
The Hyper-Scaled Investment Mandate
We must look at the mechanics of the latest funds. While I don't have the exact term of the current vehicle, the trends indicate a move towards "evergreen" structures and cross-stage flexibility. This is not a detail; it's a philosophical shift. A traditional VC fund has a 10-year lifespan. It must return money. But if you are building AI infrastructure, your runway is not 10 years. It’ might be 20 or 50. You are building what Michael Saylor would call a "monolithic" energy-asset.
Sequoia’s modified structure to allow for longer hold times and potentially index-style strategies signals they are managing a balance sheet, not a fund. This is the "algorithmic re-education" of a venture firm. And here's the contrarian angle that my experience analyzing DeFi protocols like Terra/Luna has made me hyper-sensitive to: When you alter the incentive model of the fund to match the asset, you are no longer managing risk; you are becoming the risk.
In 2020, I forked three AMMs to build localized liquidity. I quickly realized that the risk wasn't the code; the risk was the maintenance of the code. The engineering overhead kept me from understanding the user. That's what happens when the method of production (constant innovation) becomes the strategy itself. The signal is lost.
The same principle applies here. If Sequoia’s entire institutional existence becomes focused on doubling down on AI bets that require $100 million to $1 billion to be competitive, what happens to the "garage" founders? They don't go away. They just become... irrelevant to the main game. The power law is getting steeper. In the past, a $10 million seed into a company that becomes a $100 million business was a win. In the AI era, a $10 million seed is a rounding error in the model training compute.
The real insight here isn't that Sequoia is aggressive; it's that they have accepted the notion that capital itself is the most important "weight" in the model. They are pre-training the market. They aren't just financing the algorithms; they are financing the training runs that will rewrite the global economy. This is a level of power that makes even the largest hedge funds look like smallholders.
The Unsustainable Velocity of Financialized Intelligence
But we have to ask: Is this rational? Or is this the market's worst case of FOMO since the ICO mania of 2017? I'm reminded of the analogy of the "thermostat" used in system design. When a thermostat is set too high, it forces the heating unit to run continuously. A VC model that is tuned too aggressively forces companies to scale prematurely. The AI industry is the heating unit right now.
We are seeing an arms race for GPUs. The market for H100s and B100s is so constrained that the secondary market has turned into a speculative bubble. We are seeing absurd valuations for compute companies with uncertain revenue. The "art" of the deal has been replaced by the "access" to the deal. You aren't investing in a company; you're investing in a reservation queue for hardware.
This is where my 'Grounded Skeptical Mentor' persona kicks in. Based on my audit experience in Jakarta, I can tell you that when a system relies on infinite growth to remain solvent, it has a structural flaw. Terra's algorithmic stablecoin wasn't "trustless"; it was dependent on an infinite influx of capital to maintain arbitrage. Similarly, these AI valuations are currently pegged to an assumption that the cost of inference will drop exponentially, and the adoption will outpace the electricity costs. But if the compute doesn't deliver, the value has to be written down.
We must monitor not just the tech, but the energy underpinning it. This is the blind spot of the web3 community. We are so focused on the digital world of tokens and consensus that we often ignore the physical supply chain and energy requirements of the systems we rely on.
The Institutional Handshake vs. The Permissionless Handshake
When I built "BlockJakarta," I realized the true value proposition was bridging formal authority and informal agility. Regulators wanted transparency; founders want speed. The current conflict in the venture space is a scaled-up version of this. There’s the systemic aspect of this shift that we need to process.
The tension is the following: We have Sequoia, a master of the institutional handshake — access, network, distribution — pushing aggressively into a technology that is fundamentally rewriting the rules of that network. The linchpin of capitalism is trust. Historically, trust was managed by human judgment — the "good ol' boy" network, the Harvard Business School mafia, the bridge club. Sequoia is one of the last institutions where human judgment was still the primary key. Under Grady and Lin, the key is increasingly quantitative and algorithmic.
This represents a clearing event for the concept of "protocol" in the capital markets. We in the crypto world talk about "social scalability" and "trustless" consensus. We believe that math is a better arbiter than humans. Sequoia is now thinking the same way — but instead of building a decentralized protocol, they are building a centralized intelligence that aims to be the most intelligent allocator of capital on earth.
The most human thing about the old VC model was its inefficiency. The gut feelings. The golf games. The happy hours. Those were friction points. They were expensive. But they also acted as a circuit breaker. If a founder was a jerk or a technology was fragile, in many cases, the human network protected you because the network could feel it. The new model, for all its algorithmic precision, might lack those humans who can say "this doesn't feel right."
We saw what happened with Sam Altman's firing and rehiring at OpenAI. It was the messy, very human board, trying to execute "protocols" of safety over the messy, very human CEO's quest for compute. The result was a disaster. The reason that situation was so dangerous wasn't because the technology is smart; it's because the capital is overly concentrated. When one firm has the ability to dispose of a CEO, or influence the governance of the most critical company in the world, it is no longer just an investor; it is a pseudo-state.
The Crypto Bias: Web3's Response to the Centralized Compute Threat
Now, I know this is a "Crypto Briefing". We need to answer the question: What does this mean for us? It means that the long-anticipated fight between crypto and AI isn't about Dogecoin predictions. It's about sovereignty.
We have a thesis: Education is the new mining rig for the mind. If we don't understand the architecture of this new financial intelligence, we will be subservient to it. The people who understand how to prompt the models, how to own the compute, and how to validate the data are the new capitalists.
But there is hope in a decentralized methodology. This is the "Contrarian" part.
The contrarian angle is that the "open-source" strategies of decentralized compute projects are actually more resilient than Sequoia's massive checkbook approach.
Why? Because the cost of access isn't the only variable. The cost of governance is higher for a centralized system than a decentralized one. When you control trillions of dollars in compute, you have a massive liability. You have to secure that compute from cyber attacks, geopolitical shifts, and energy price fluctuations.
But if you have permissionless access to a global network of smaller, distributed compute nodes (the promise of the web3 GPU network), you are buying energy in a fragmented, more resistant way. You are buying "optionality."
I remember dissecting the failure of "Trustless" systems for my 50-page Terra analysis. They fail when the incentives converge. If Sequoia builds one monolithic and concentrated cluster, the entire system is a target. One electromagnetic pulse, one skilled hacker, or one nation-state with a grudge can take it down.
In the web3/GPU sharing economy, the target is diffuse. It’s harder to kill thousands of distributed GPUs in Indonesia, Iceland, and Brazil than it is to kill one data center in Silicon Valley. This is the true strength of decentralization: not speed, but resilience. In the AI era, resilience is the new alpha.
The Social Engineering of the Infinite Loop
When the market sleeps, the architects wake up. And Sequoia is wide awake. This "Anthropological Identity Observer" in me sees the venture capital space as a tribe. And tribes need creation myths. The myth of the "eat what you kill" founder is being replaced by the myth of the "scalable intelligence" — the idea that the best investment is one that minimizes human error. But the irony is that an investment process designed to minimize human error is itself a product of human bias — the bias of collective FOMO.
Institutional investors are herding into AI faster than they did into crypto. Why? Because the narrative is clearer. You can see a tangible product like ChatGPT. You can see a tangible asset like an Nvidia GPU. It feels real.
But it feels more real than a protocol because it is more physical. And that physicality is what makes it dangerous. You cannot burn a Bitcoin to make a fire. But you can burn a GPU. You can overhead a data center. You can run out of water.
The central thesis of Lucas here is that Sequoia is not "alpha." Sequoia is "beta" — the benchmark. They are the market. And by going all in on AI, they are creating a feedback loop that forces every LP, every founder, and every competitor to follow suit. This is not independent innovation; this is institutionalized herding on a massive scale.
My advice for the mature investor in this space is to not treat Sequoia as a leader, but as a lagging indicator. By the time they are this aggressive, the arbitrage is gone. The real alpha is in the boring, difficult work of training human beings to understand what these machines can and cannot do. It's in the legal frameworks and the compliance structures. It's in the "blocks" of the value chain that are not just the GPU, but the instruction manual, the regulatory clearance, and the ethical standard.
The Regulated Revolution
The last point on the future of Venture Capital and AI: Regulation. Sequoia is not an anarchist. They are the ultimate insider. They will be compliant. That's why their aggressive AI pushes will likely shape regulation even more than lobbying. In the same way that I wrote about post-ETF approval, we are about to see financial instruments that securitize compute power, future energy output, and inference rights.
The takeaway is: capitalism has found a new frontier, and its name is not 'crypto' or 'AI' solely; it is 'Attention' — but more importantly, 'Compute'.
We must bridge the gap. In the crypto world, we have a toolset for this: Token incentives. If Sequoia is buying the machinery, we should be buying the social consensus around that machinery. The future isn't about owning the GPU; it's about owning the right to exist on that GPU. That is a status game. It’s a game you play with code, not with checkbooks.
The pressure we feel in this bull market is the tug of war between the old means of production (centralized capital) and the new means of production (algorithmic scarcity). The valuations are high because the belief is high, but the collateral could evaporate if the power grids fail.
As we look towards 2025 and 2026, the signals read differently. The most important architectural shift isn't the model or the chip; it’s the foundation of the trust.
Will we simply substitute the old intermediaries of finance with new intermediaries of compute? Sequoia is betting on the latter, trading the social contract for a data contract. But in the ethereum core dev trenches, we learned a simpler truth: Trustless systems don't mean you don't trust anyone; it means you don't have to rely on them. And in a world where the machines are taking over the spreads, the last bastion of human value is not the code, but the verification of the human behind the code.
This is why we build. This is why we learn. Not to chase the machine, but to put it in its box. The question is whether Sequoia is building a box, or being built by the machine. Watch the capital, and you'll see the answer.
