July 2025. TCS India announces twelve thousand job cuts. CEO Krithivasan is explicit: "This is not because of AI giving some 20% productivity gains." Ten months later, on May 19, 2026, Standard Chartered announces between seven and seven thousand eight hundred positions eliminated by 2030. CEO Bill Winters attributes the cuts directly to artificial intelligence. Group CIO Veiga leaves no room for interpretation: "Job role reductions in favor of the machines, and that will accelerate." Same structural operation. Opposite narratives. Ten months apart.

The question organizing this article is not whether displacement is occurring. That was documented in previous installments of this series. The question is different: what changed between July 2025 and May 2026 for corporate discourse to reverse direction?

This is the seventh article in the UBI vs Luddite Manifesto series. Previous ones mapped geographies. This one maps timelines. The question is not where displacement occurs, but in what order.

The year 2025 built a trend worth reading carefully. February: DBS Singapore announces four thousand positions over three years, first global bank to explicitly tie AI to a concrete timeframe. That same month, Commerzbank adds three thousand nine hundred positions and six hundred million euros in AI investment for the 2026-2030 period. May: Klarna admits that its AI-centric approach reduced customer service quality and rehires humans. September: Bosch announces thirteen thousand positions by 2030 at its Mobility plants in Germany. November: Clifford Chance London eliminates ten percent of its support services with explicit attribution to AI plus offshoring to India and Poland. Also in November, Brynjolfsson of Stanford publishes Canaries in the Coal Mine, confirming the hollowing out of the experience curve in junior profiles.

What 2025 showed was not just the volume of cuts. It was the distribution of language. Companies that cut: many. Companies that said "it's because of AI": few. TCS, Oracle, Cognizant, Wipro and Accenture avoided naming the technology. Top-tier global banks that tied specific numbers, timelines and attribution to AI: none throughout 2025.

That changed five months after year-end.

Between January and May 2026 the record shifts tone. The ILO World Employment and Social Outlook 2026 report incorporates "AI hesitation in hiring" as an explanatory variable in its models for the first time. Acemoglu, who in previous years published macroeconomic analyses on automation, pivots toward a more specific concept: Knowledge Collapse, the deterioration of cognitive capabilities in organizations that outsource reasoning to AI systems. In April, Claude Opus 4.7 launches. Yann LeCun, from X, reiterates that AGI doesn't come from LLMs. In May, Commerzbank adds three thousand additional positions to those already committed.

On May 19, Standard Chartered becomes the first top-tier global bank to announce cuts with the complete format: numbers, timeline, attribution to AI. The stock rose two and a half percent that day. The market didn't read the announcement as an alarm signal. It read it as efficiency.

Here appears the asymmetry that organizes everything else.

There are three curves moving at different speeds, and in that lag lies the relevant information. The corporate announcement curve has dates: Bosch 2030, Standard Chartered 2030, DBS 2028, Commerzbank 2030. The verifiable total of positions committed in those announcements hovers around twenty-nine thousand. Bloomberg Intelligence, in a survey of CIOs from ninety-two banks, projects around two hundred thousand global banking positions over three to five years. Morgan Stanley estimates two hundred thousand European banking positions by 2030. Goldman Sachs reports sixteen thousand net positions eliminated per month in the U.S. The corporate curve has a calendar.

The technical curve doesn't. OpenAI hasn't announced GPT-6. Anthropic hasn't announced Claude 5. Google hasn't announced Gemini 4. Meta hasn't announced Llama 5. Companies producing the models don't publicly commit to delivering the successor by 2030. Companies using the models do commit to cutting by 2030. That asymmetry isn't a technical detail. It's the architecture of the problem.

The physical curve has bottlenecks that also don't appear in headlines. TSMC packaging is exhausted. In the U.S., around twenty-five data centers were canceled in 2025, versus six in 2024. The electrical connection queue in ERCOT Texas grew three hundred percent. Wait times to connect infrastructure to the power grid range, depending on region, from two to five years. The energy needed to execute the promise isn't available within the timeframe corporate announcements assume.

Organizations are making restructuring decisions based on projections of technology that doesn't yet exist in the form needed. The public scale of commitment and precision of timeline are novel; the dynamic isn't. How far these announcements will go before hitting hard limits is something that still isn't clear.

Institutional forecasts function as mirrors, not predictions. The WEF Future of Jobs 2025 projects ninety-two million jobs displaced and one hundred seventy million created by 2030, with a positive balance of seventy-eight million. McKinsey, in November 2025, reports that thirty-two percent of surveyed companies expect workforce reduction. The OECD estimates that thirty-nine percent of current skill sets will be obsolete by 2030. These figures aren't verifiable predictions. They're signals of what institutional actors are willing to say out loud.

What can be read, with available data, is a probable sequence for the next twenty-four months. This is the author's reading, not consensus or announcement. In the next six to twelve months, two to four additional global banks will replicate the Standard Chartered format: numbers, timeline and explicit AI attribution. In the following twelve to eighteen months, one of the Big Four or Big Three consultancies will formalize the same scheme. Between eighteen and twenty-four months, a top-six Indian BPO firm will explicitly name AI, closing the silence cycle that TCS opened in July 2025.

Numbers and timelines are moving faster than the technology that justifies them.

This prediction has conditions that would falsify it. Three or more Klarna-type reversals at top-tier companies during 2026-2027 would change the discourse direction. Effective enforcement of the European AI Act after August 2026, with real sanctions and not just declarations, would alter incentives. A capital investment cut at major cloud providers during 2026 would halt part of the infrastructure supporting the promises. A major sectoral strike attributable to AI during the same period would rewrite corporate communication policy. A non-LLM technical breakthrough with capacity gains at much lower compute would change the entire technical curve.

What cannot be predicted is different from what cannot be observed. We cannot predict when the model becomes politically unsustainable, when a coordinated political response emerges, when the hype cycle breaks, or when the physical bottleneck becomes structurally blocking for the entire industry.

On January 30, 2026, SpaceX filed with the FCC a plan to deploy one million satellites in low orbit as distributed processing centers. Projected capacity: one hundred GW of AI compute, equivalent to twenty percent of current U.S. electrical consumption. The document uses language worth quoting verbatim: this system is "the first step toward becoming a Kardashev II civilization — one that can harness the sun's full power." xAI was acquired by SpaceX. The combined valuation reaches 1.25 trillion. Musk declares that cheaper AI compute could be in space within two to three years.

When terrestrial bottlenecks limit the displacement timeline, actors who promised displacement don't accept slowdowns. They propose next-order-of-magnitude infrastructure. If earth won't provide the energy, space will. If the power grid isn't enough, satellites. Corporate cutting timelines don't adjust to bottlenecks. Bottlenecks are declared obsolete.

The next chapter in this series will address responses. This chapter doesn't propose. It only predicts. People whose positions are on those corporate calendars exist today, while the satellites haven't been launched yet.

How much longer can corporate calendars run ahead of the infrastructure supposedly sustaining them?

Sources:

1. Standard Chartered Group, Q1 2026 earnings release, May 19, 2026 — workforce reduction announcement with explicit AI attribution by CEO Bill Winters and Group CIO Veiga.

2. Brynjolfsson, E. et al., Canaries in the Coal Mine, Stanford Digital Economy Lab, November 2025 — empirical confirmation of experience curve hollowing in junior profiles.

3. WEF, Future of Jobs Report 2025 — job displacement and creation projections toward 2030 (92M displaced / 170M created).

4. ILO, World Employment and Social Outlook 2026, January 2026 — incorporation of "AI hesitation in hiring" as explanatory variable.

5. SpaceX, FCC filing for low-orbit AI compute satellite network, January 30, 2026 — projected 100 GW capacity and Kardashev II civilization reference.