AI will hollow out the City but must be embraced, says tech investor - The Observer

TEXT START: The UK’s financial services sector, which contributes 8% of economic output, will be “massively commoditised” by AI with drastic consequences for the banking workforce, according to a former Sequoia partner who has now raised $970m across three AI investment funds in London.

The Dissection

This is a transition memo disguised as news. It accurately reports that AI is attacking software development, analysis, investment research, strategy, and thought leadership—the cognitive layers that justify much of the City’s headcount. It then launders the consequences through the word “embrace.”

The investor is not merely describing disruption. He is normalizing labor liquidation while lobbying for the conditions that accelerate it: liquid labor markets, weaker non-competes, and fewer institutional restraints on AI startups. The article treats the survival of financial services as if it implied the survival of financial workers. It does not. The sector can preserve revenue, output, and executive compensation while discarding the workforce that once produced them.

The Core Fallacy

The central error is the historical analogy that displaced workers will simply “elevate to another kind of role.” Earlier technological waves created new sectors that could absorb labor. Agentic AI attacks cognitive work across sectors simultaneously, at machine speed and with compounding cost advantages. There is no demonstrated human-only frontier large enough to absorb millions of displaced analysts, developers, researchers, administrators, and support staff.

The claim that there is “nothing you can do but embrace it” confuses inevitability of technical competition with inevitability of the resulting ownership structure. Workers can be displaced even while the gains are captured by a narrow class of AI-capital owners. Removing non-competes may help investors recruit talent; it does not restore mass productive participation.

Under the Discontinuity Thesis, the reported 30%–50% task automation is already sufficient to begin the kill sequence: cognitive automation dominance, institutional inability to preserve human-only domains, then collapse of access to economically necessary labor. Full automation is not required. Marginal labor demand only has to fall far enough to break the wage-to-consumption circuit.

Hidden Assumptions

• New AI-created roles will appear at sufficient scale and quickly enough to absorb those removed from existing roles.
• Human oversight, accountability, and trust will remain labor-intensive rather than becoming software features or legal fictions.
• Short-term bank profits prove durable institutional strength rather than reflecting a lag before automation reaches the balance sheet.
• A more liquid labor market benefits displaced workers rather than making them easier to replace and cheaper to hire.
• State-backed AI funds will produce broad prosperity rather than socialize risk while concentrating ownership.
• The public can continue consuming after wage income collapses through some unspecified mechanism.
• Workers can “embrace” AI without owning or controlling the systems that replace them.
• The City’s universities, openness, and startup ecosystem will convert Britain into a sovereign AI power rather than a convenient talent and capital basin for foreign or concentrated owners.

Social Function

Classification: partial truth, transition management, elite self-exoneration, and ideological anesthetic.

The article admits enough destruction to appear candid, then strips the destruction of political meaning. “Embrace” converts a distributional conflict into a personal attitude problem. The investor’s message is: accept your replacement, deregulate the labor market, subsidize the firms doing the replacing, and call the resulting concentration innovation.

The non-compete dispute is real but secondary. It determines who captures the transition among competing firms; it does not determine whether mass labor remains necessary. The deeper function of the article is to prepare the public for a City that survives as an AI-managed capital machine while its employment base is treated as obsolete inventory.

The Verdict

The article identifies the City as an early casualty of the discontinuity but understates the systemic consequence. Finance will likely survive in a more automated, concentrated form. The banking workforce will not survive at its present scale, and the missing jobs will not automatically reappear elsewhere.

This is not creative destruction in the old sense. It is productive participation being severed from ownership. The investor calls that process unavoidable because it benefits the class positioned to own the replacement machinery. The City is not being saved. It is being hollowed out, monetized, and presented as progress while the wage-consumption circuit is dismantled underneath it.