In the liquidity fog of 2017, I watched ICO whitepapers promise decentralised utopias while their presale allocations were engineered to dump on retail. The patterns were obvious if you looked at the incentives, not the hype. Today, a similar structural shift is unfolding not in crypto, but in the AI investment landscape. SoftBank, the Japanese conglomerate that once bet billions on WeWork and Uber, just appointed Yossi Cohen—the former director of Mossad—as a strategic advisor for its AI investments. The market reaction was muted. The deeper implication is not.
This is not a routine hire. Cohen is not a tech executive, a venture capitalist, or a machine learning researcher. He is a career intelligence officer who ran Israel's foreign espionage agency for five years, overseeing operations that remain classified. His appointment signals something far more radical than a simple 'security consultant' role. SoftBank is publicly acknowledging that the highest-stakes competition in AI is no longer just about model accuracy or compute scale—it is about geopolitical intelligence, supply chain control, and the ability to navigate the shadow war between states. For a firm that holds a controlling stake in Arm, the architecture behind the majority of AI chips, this move transforms SoftBank from a passive capital allocator into an active participant in the global security architecture.
To understand the magnitude, we need to map the context. SoftBank's Vision Fund, once the largest technology investment vehicle in history with nearly $100 billion under management, has been retrenching after the WeWork debacle and the broader tech correction. Founder Masayoshi Son has repeatedly stated that he is 'all in on AI' and that Artificial General Intelligence (AGI) will arrive within a decade. The firm's crown jewel is Arm, which designs the instruction set architecture used by nearly every smartphone and an increasing share of AI servers. Arm's IPO in 2023 was a landmark event, cementing SoftBank's leverage over the AI hardware stack. But Son understands that controlling the chip layer is not enough. The true bottleneck in the AI era is not silicon—it is trust, security, and access to sensitive talent pools. Cohen's appointment directly addresses this bottleneck.
Let me peel back the layers, because this is where the structuralist analysis kicks in. The first layer is capital deployment. SoftBank’s investment thesis has historically been growth-at-all-costs, funding companies that disrupt incumbents. That thesis is now being augmented with a 'security-first' filter. Cohen’s role is likely to assess the geopolitical risk profile of potential investments—especially those involving national security, dual-use technologies, and cross-border data flows. This is not just about avoiding bad bets; it is about opening channels that no other VC can access. Israel is a global hub for cybersecurity, AI-driven surveillance, and military-grade autonomy. The Israeli tech ecosystem is deeply intertwined with Unit 8200, the intelligence unit that spawned many of the country's most successful startups. Cohen, as former Mossad chief, has a Rolodex that includes the founders of those startups, the generals who oversee their contracts, and the regulators who approve their exports. This is a distribution advantage that no amount of capital can replicate.
Yields are just risk wearing a disguise. In the crypto world, we learned that high-yield DeFi protocols are often just high-risk lending chains disguised as innovation. The same principle applies to AI venture capital. The highest-return AI investments are likely to be in areas that are politically sensitive, regulated, or outright controversial—think facial recognition, autonomous weapons, and predictive policing. SoftBank, by hiring Cohen, is signaling that it is willing to take on that risk, but only if it can manage it with intelligence-grade diligence. The corollary is that SoftBank will likely avoid investments in markets where Cohen's background creates a liability, such as China, where any association with Mossad is a red flag for regulators. This is a strategic narrowing of the investment universe, but it comes with a higher potential payoff per deal.
The second layer is competitive positioning. The global AI investment landscape is fragmented. On one side, you have US-based venture firms like a16z, Sequoia, and General Catalyst, which have deep networks in Silicon Valley but limited access to national security-focused startups. On the other side, you have sovereign wealth funds and state-backed entities that prioritize strategic control over financial returns. SoftBank, with its Japanese base, Middle Eastern LP capital (the Public Investment Fund of Saudi Arabia is a major investor in Vision Fund), and now an Israeli intelligence advisor, occupies a unique middle ground. It can source deals from the Israeli security complex, fund them with Saudi and Emirati capital, and deploy them globally using Arm's design ecosystem. This triangular relationship is unprecedented. It also carries significant execution risk, as the political tensions between Israel and the Gulf states could flare up at any moment.
Systemic rot is hidden in the fine print. The fine print here is that Cohen's appointment is not just about deal sourcing—it is about due diligence. In AI, technical due diligence is notoriously difficult because the field moves too fast and proprietary models are often black boxes. A venture fund cannot audit a GPT-4 competitor the same way an auditor checks a bank's balance sheet. But someone with a background in intelligence tradecraft can assess the trustworthiness of the founding team, the resilience of the supply chain, and the likelihood of regulatory backlash. This is a form of risk assessment that traditional VCs lack. However, it also introduces a new class of risk: the risk of being perceived as a tool of a foreign intelligence service. Some AI startups may refuse SoftBank's money for fear of being associated with Mossad, especially in Europe and parts of Asia. The very advantage that Cohen brings could become a liability.
Let me now pivot to the contrarian angle. The prevailing narrative in the tech press is that this is a smart move to enhance SoftBank's security expertise. I disagree. The real story is that SoftBank is betting on a decoupling thesis: the idea that the AI industry will bifurcate into two camps—one aligned with US/Israel security interests, and one aligned with China. By hiring Cohen, SoftBank is publicly choosing sides. This is not a neutral advisory role; it is a political statement. The contrarian view is that this decoupling is not inevitable, and that SoftBank's move may accelerate the very fragmentation it seeks to exploit. In a decoupled world, the value of intelligence networks increases, but the liquidity of the overall market decreases. SoftBank may find itself trapped in a high-risk, high-return portfolio that is difficult to exit when sentiment shifts.
Correlation is the siren song of fools. While the market is currently pricing AI stocks as a monolithic 'growth' sector, the addition of geopolitical risk differentiation will create divergence. Companies with strong ties to Western security apparatuses will trade at a premium; those with exposure to Chinese or Russian markets will face a discount. SoftBank is positioning itself to capture that premium, but it is also making a leveraged bet that the current geopolitical alignment persists. If the US-Israel relationship sours or if the Middle East peace process changes, the entire thesis crumbles.
Now, how does this connect to the crypto world? As a cross-border payment researcher, I see a direct parallel. The evolution of stablecoins and payment rails is also being shaped by geopolitical forces. USDT dominates the stablecoin market, but its reserves have never been fully audited, and Tether's compliance with sanctions is opaque. The same 'trust but verify' dynamic applies. When a former intelligence chief joins a major financial institution, it signals that the institution is willing to operate in the gray zone between regulated finance and state influence. In crypto, we already see this with funds that are backed by sovereign wealth funds from authoritarian states. The line between legitimate investment and statecraft is blurring. SoftBank's move is a harbinger of a broader trend: the weaponization of capital flows in the AI and crypto sectors.
Volatility is the tax on certainty. The market's muted reaction to Cohen's appointment is a sign of complacency. Investors are not pricing in the tail risks of a scandal, a regulatory probe, or a geopolitical backlash. But history tells us that when intelligence operatives enter the private sector, the volatility tends to follow. Consider the case of Blackstone's hiring of former CIA directors, or the revolving door between the NSA and Palantir. These moves often precede major regulatory crackdowns or public controversies. The same will likely happen with SoftBank. The tax on certainty—the premium that investors will demand for holding SoftBank's equity or its portfolio companies—will increase over time. For now, the market is ignoring it, but that is precisely when the risk is most underpriced.
Let me ground this in my own experience. During the 2022 crash, I analyzed the systemic contagion from Terra/Luna and Celsius. The common thread was not fraud, but the failure of risk management in a highly correlated environment. Everyone assumed that yields were independent, but they were all tied to the same liquidity pool. Today, the AI investment landscape is similarly correlated. Everyone is chasing the same few large language models, the same GPU supply chains, and the same regulatory frameworks. By hiring Cohen, SoftBank is trying to break that correlation by introducing a novel risk factor: geopolitical intelligence. But this new factor introduces its own correlations. If SoftBank's intelligence-driven bets fail, the contagion will not be limited to its portfolio; it will spill over into the broader AI ecosystem, affecting valuations across the board.
Innovation often precedes regulation by a decade. The Cohen appointment is a bet that regulatory frameworks will eventually align with the security-focused approach. But regulators are lagging. In the US, the CFIUS (Committee on Foreign Investment) is scrutinizing foreign investments in AI startups. The EU is drafting the AI Act. China has its own data security laws. SoftBank is trying to get ahead of this regulatory curve by embedding security expertise at the top. But the risk is that regulators may view this as an attempt to bypass rules rather than comply with them. The next decade will see a regulatory pendulum swing, and SoftBank's position may be too exposed.
So what is the takeaway? For investors in both AI and crypto, the message is clear: the era of apolitical, pure-technology investing is over. Capital flows are now governed by geopolitical risk, and the firms that can navigate that risk will capture outsized returns. SoftBank is making a bold bet that its intelligence network will give it an edge. But in doing so, it is also painting a target on its back. The next black swan may not come from a technological failure, but from a geopolitical event that exposes the fragility of this new, shadowy infrastructure.
History doesn’t repeat, but it rhymes in code. In 2017, I saw the same pattern in ICOs: a small group of insiders with privileged information would exit before the crowd. In 2025, the pattern is repeating in AI, with intelligence networks replacing insider trading. The code is different, but the incentives are the same. Those who can read the shadows will survive. Those who chase the light will be left holding the bag.
SoftBank's shadow state is now operational. The question is whether it will illuminate the path to AGI or cast a long, dark shadow over the entire industry.