The code compiles, but does it heal?
In the last quarter, Sequoia Capital deployed over $1.2 billion into AI startups, a figure that surpasses its entire crypto portfolio allocation in 2021. Under the new leadership of Roelof Botha and the aggressive deal-making of partners like Alfred Lin and Ravi Gupta, the firm has become the single largest private investor in large language models and AI infrastructure. Yet, the silence around the structural risks of this concentration is deafening. It reminds me of the silence I felt in May 2022, when Terra's collapse was imminent, and no one wanted to speak the truth. Silence is the loudest indicator of systemic rot.
Context: The New Sequoia and the Old Playbook
Sequoia has always been a trendsetter, but its pivot under Lin and Grady (the new generation of partners) is different. Historically, the firm’s investments were diversified across enterprise, consumer, and fintech. Now, nearly 40% of its active investments are in AI-related companies, from foundational models like OpenAI and Anthropic to application layers like Cursor and Harvey. This mirrors the pattern I observed in 2017 during the ICO boom, when VCs suddenly shifted entire funds into token sales. Back then, I refused to pitch technical whitepapers to venture capitalists. Instead, I spent three months writing a 40-page manifesto titled "The Moral Architecture of Trust," analyzing the ethical implications of smart contracts versus traditional banking. That document received 12 substantive replies from academics who valued the ethical framework over financial yield. Today, I see a similar moral vacuum in the AI rush.
The narrative is seductive: AI will transform every industry, and Sequoia is simply placing the largest bets. But beneath the surface, the firm is manufacturing a new type of scarcity—not of tokens, but of compute and talent. By funding competing AI labs, Sequoia creates a race where only the best-funded survive, inflating valuations to levels that even the most optimistic public markets struggle to justify. This is not innovation; it is centralization through capital.
Core: The Architecture of Trust and the Fragility of Centralized AI
Let me dissect three structural flaws in Sequoia’s aggressive AI strategy, drawing from my experience auditing DeFi protocols and analyzing Layer2 sequencers.
1. The Liquidity Fragmentation of Talent
In DeFi, we often hear about "liquidity fragmentation" as a problem that VCs push new products to solve. But I've always argued that liquidity fragmentation isn't a real problem—it's a manufactured narrative. Similarly, Sequoia's AI investments are fragmenting the talent pool. They fund multiple startups in the same vertical, forcing engineers to compete for the same pool of PhDs. This drives up salaries and burn rates, creating a dependency on continuous fundraising. The result is a system where startups are not building for sustainability but for the next round. I've seen this pattern in crypto: projects that raise $100M at a $1B valuation but have no product-market fit. The code compiles, but does it heal?
2. The Centralization of Sequencers
In Layer2 scaling, the dirty secret is that sequencers are essentially single centralized nodes. "Decentralized sequencing" has been a PowerPoint slide for two years. Sequoia’s AI investments suffer from the same problem. They invest in proprietary models that are gated by centralized APIs. The infrastructure (GPUs, cloud, data centers) is controlled by a handful of companies like Nvidia and AWS. Sequoia’s portfolio companies, no matter how innovative, are tenants on someone else's land. This is not the permissionless innovation that blockchain promises. Trust is not encrypted; it is woven. And the weave here is thin—controlled by a few data centers and a few VCs.
3. The Ethics of Autonomy
Based on my experience launching the "Conscious Algorithms" salon series in 2025, where I brought together philosophers, AI ethicists, and blockchain developers, I’ve learned that the true test of any technology is its ability to grant autonomy to the user. Sequoia’s AI investments are building tools that, at best, augment human productivity. But at worst, they create dependence. When I interviewed a developer at a Sequoia-backed AI coding assistant, he admitted that the platform’s license terms allow the firm to use all code written on it to train their models. This is the same exploitative model that Web3 aims to dismantle. The silence around this is chilling.
Furthermore, the aggressive AI investments are reshaping venture capital norms. High valuations are sustained not by revenue but by the fear of missing out. I recall the Terra/Luna collapse in 2022, when I withdrew from social media for six weeks to document 14 case studies of financial trauma. The pattern is identical: a narrative of inevitability, a rush of capital, and a disregard for fundamental risk. The only difference is the asset class.
Contrarian: The Blind Spot of Scale
One might argue that Sequoia’s aggressive AI strategy is rational because the technology is transformative. But the contrarian angle is that scale itself is the enemy of resilience. In decentralized systems, we value many small nodes over one giant node. Sequoia is creating a giant node of capital, and any failure in that node—a single bad investment, a regulatory crackdown, a shift in public sentiment—could cascade. The firm’s portfolio is so concentrated that a downturn in AI sentiment could wipe out billions.
Moreover, the high valuations mask technical flaws. For example, many Sequoia-backed AI companies claim to have proprietary data moats, but in reality, they are training on publicly available data. The barrier to entry is lower than they admit. The same was true in crypto: projects with $100M valuations had 10 lines of code. Feminine wisdom asks not 'how fast?' but 'how whole?' And Sequoia's portfolio is not whole—it is fragmented, over-leveraged, and vulnerable.
There is also a cultural blind spot. Sequoia’s leadership, under Lin and Grady, is known for a "founder-first" ethos. But in practice, this means prioritizing the interests of the VC over the community. In my mentorship program "Women of the Chain," I saw how homogenous decision-making leads to blind spots. The AI industry is even more male-dominated than crypto. The lack of diverse perspectives means that ethical considerations are often an afterthought. The code compiles, but does it heal?
Takeaway: The Future of Venture Capital and the Soul of the Industry
The aggressive AI investing by Sequoia is not just a financial strategy; it is a philosophical statement. It says that capital concentration is the most efficient way to drive innovation. But I’ve spent 29 years in this industry, and I’ve learned that efficiency without resilience is a mirage. The blockchain community knows this: we build trust through decentralization, not through central planning.
As we move forward, I challenge every founder and investor to ask: Who wrote the rules of this AI race? And who broke them? The silence of the current euphoria will be broken by the next crash. Until then, I will keep writing, keep questioning, and keep weaving a different narrative. One where trust is not encrypted, but woven. Where the code compiles, and it heals.