The Pivot After the Panic: How Silicon Valley Learned to Stop Worrying About AI Jobs

In early 2023, the forecasts were dire. Goldman Sachs published research suggesting that artificial intelligence could replace the equivalent of 300 million full-time jobs globally. The CEO of an influential chatbot company told a Senate subcommittee that the technology posed a fundamental threat to democratic participation. A prominent robotics entrepreneur declared that his industry's trajectory would make human labor "optional" within two decades. The language was not cautious. The timelines were not hedged.

By mid-2026, that rhetorical heat has substantially cooled. Jensen Huang, whose company's processors sit at the center of the AI buildout, told investors and press that previous "doomsday" predictions had overstated the near-term effects. His counterpart at a leading AI laboratory echoed the reassessment. The industry that spent two years telling Washington that existential risk demanded immediate regulatory attention is now telling the same audience that job displacement is more gradual, more nuanced, and less urgent than advertised.

The pivot is real. It is also instructive.

What the Industry Said Then

The 2022 and 2023 period produced a concentrated series of high-profile warnings about AI's labor effects. Goldman Sachs's March 2023 report, widely cited across policy circles, estimated that roughly two-thirds of jobs in the United States and Europe faced some degree of AI exposure, with around a quarter of current tasks potentially automatable. The Oxford Martin Programme on Technology and Employment published estimates in the same window suggesting that up to 47 percent of U.S. jobs were at risk of computerization within two decades. These figures circulated through Congressional testimony, regulatory consultations, and media coverage with minimal contestation.

The framing carried a specific character. AI was not merely another automation wave, in this telling — not the displacement of specific manufacturing tasks or the automation of particular clerical routines. It was different in kind. The technology's capacity to generate content, reason through novel problems, and perform cognitive labor meant that the traditional safety valves — retraining, new sector creation, productivity gains enabling new employment — might not operate as they had in prior transitions. The World Economic Forum's 2023 Future of Jobs report reinforced this, projecting that AI would displace 85 million jobs globally by 2025 while creating 97 million new ones — a net positive over that specific horizon, but one that required substantial workforce adaptation.

What these projections shared was a temporal compression. The displacement, when it came, would arrive faster than historical precedent suggested. The transition would be concentrated in knowledge-economy roles previously considered insulated: legal research, financial analysis, software development, journalism, medical diagnostics. The political economy implications — tax base erosion, social welfare strain, democratic legitimacy questions — were presented as urgent and near-term.

What the Industry Says Now

France24 reported on 28 May 2026 that AI leaders including Jensen Huang and Sam Altman are moderating those earlier warnings, suggesting that the "doomsday" framing overestimated near-term disruption. The language shift is notable in its specifics. Huang has spoken in recent months about AI as a productivity multiplier rather than a labor substitute — a tool that makes individual workers more capable rather than rendering them redundant. Altman has similarly emphasized the technology's role in creating new categories of work and enabling human creativity rather than suppressing it.

The recalibration has not been uniform, and the sources do not specify precisely when each executive changed their public framing or what internal data prompted the shift. But the direction is consistent enough to constitute a pattern. The executives who told Washington in 2023 that AI required immediate, fundamental regulatory attention because of labor displacement risks are now, in 2026, telling the same audiences that the transition will be gradual enough to manage through existing policy mechanisms.

Nvidia's investment posture provides a structural counterpoint to the softer labor rhetoric. As of May 2026, Huang announced plans for approximately $150 billion in annual investment in Taiwan, reflecting the company's assessment that AI infrastructure demand will remain robust for the foreseeable future. That investment — concentrated in semiconductor manufacturing and the broader supply chain — is the kind of capital expenditure that shapes labor markets over decades, not quarters. It creates skilled manufacturing roles in one geography while simultaneously accelerating the automation capacity that supposedly won't displace many jobs.

The Structural Logic of Industry Positioning

The pattern becomes easier to understand when placed against the legislative timeline. In 2023 and 2024, AI companies faced a moment of genuine regulatory uncertainty. The European Union was finalizing its AI Act. The U.S. Congress was debating multiple AI governance bills. The question of how — and by whom — the technology would be governed was genuinely open.

In that environment, the labor-displacement argument served a specific function. It redirected regulatory attention toward workforce adaptation and social safety net questions rather than toward the more immediate concerns of the companies building the technology: compute access, data rights, intellectual property liability, and cross-border data flows. Framing AI as a labor shock rather than a concentration-of-power problem made the technology's risks feel diffuse and political rather than concentrated and technical. Workforce disruption is a problem that governments can address through training programs and extended unemployment benefits. It is not a problem that requires restructuring the business models of the companies producing the underlying technology.

The industry also benefited from the existential-risk framing in a different register. If AI posed a credible threat to human civilization — as several prominent laboratory leaders argued in Senate testimony and public statements — then the appropriate policy response was not the incremental, sector-specific regulation then circulating in European and American proposals. It was something closer to an Manhattan Project model: concentrated AI development under state partnership, with the leading private firms as indispensable partners. That framing positioned the companies themselves as both the source of the threat and the necessary partners in its management.

The softening of labor warnings in 2026 tracks against a changed legislative environment. The most aggressive U.S. AI legislation proposals have largely stalled. The EU AI Act's implementation is proceeding on a timeline that gives major providers operational space. The immediate regulatory crisis the industry seemed to face in 2023 has not materialized in the form that observers predicted. When the urgency dissipates, the rhetoric adjusts.

What the Evidence Actually Shows

Separating the political economy of messaging from the underlying empirical picture requires examining what labor market data has actually shown since the generative AI surge began in late 2022.

The evidence is messier than either the 2023 alarmists or the 2026 minimizers acknowledge. In the United States, the technology sector has experienced significant layoffs since 2022 — tens of thousands of positions eliminated at major firms — but attributing these specifically to AI-driven automation versus broader cyclical correction or pandemic-era over-hiring remains methodologically contested. The Economic Policy Institute and other labor-focused research organizations have documented displacement in specific roles: entry-level software testing, basic content moderation, standardized legal document review. These are real effects on real workers.

Globally, the pattern is more uneven. Nations in earlier stages of manufacturing expansion — particularly in Southeast Asia and parts of Africa — face different automation exposure profiles than post-industrial economies. The International Monetary Fund's 2024 labor market analysis suggested that AI could affect around 40 percent of jobs in advanced economies but a lower percentage in developing economies, where formal cognitive labor roles constitute a smaller share of total employment.

The sources do not permit a definitive settlement of the displacement question. What can be said is that the wholesale labor-market transformation predicted for the 2020s has not materialized at the pace and scale advertised in 2023. The transformation is ongoing; its ultimate scope remains genuinely uncertain.

The Stakes Going Forward

The recalibration in executive rhetoric carries real consequences beyond the messaging sphere. If policymakers internalize the softer framing — that AI-driven job displacement is gradual, manageable, and net-positive over relevant horizons — they are less likely to pursue the more aggressive interventions that labor advocates have been requesting: universal basic income pilots, substantial severance and transition funds for displaced knowledge workers, corporate accountability mechanisms for the downstream labor effects of automation decisions.

The companies benefiting from the changed framing are precisely those best positioned to shape it. Nvidia's Taiwan investment trajectory signals a long-term infrastructure commitment that will influence labor markets across multiple sectors and geographies. The firms controlling AI development are also the firms with the deepest pockets for public relations, the most sophisticated understanding of how Washington works, and the most direct relationships with the policy makers who will determine the regulatory environment for the next decade.

There is nothing unusual in an industry moderating its public warnings once the regulatory moment has passed. But the AI labor question is not simply a messaging contest between industry and regulators. It is a question about how productivity gains from a transformative technology will be distributed. If the gains accrue primarily to capital — as historical automation waves have tended to do — while the disruption falls primarily on labor, the policy choices made in the next few years will matter enormously. The framing matters because the framing shapes what policymakers believe is necessary.

What the sources suggest, taken together, is that the companies best positioned to know what AI will actually do to labor markets have an interest in a particular answer to that question. Whether that answer is right — whether the displacement genuinely will be gradual, manageable, and net-positive — is a question that will be answered by history, not by press releases.

Monexus covered the 2023-2024 AI labor warnings as they circulated in policy and media discourse. The France24 reporting on the current recalibration reflects a consistent editorial approach: tracking what the industry's most prominent voices say, and noting when the framing shifts against the weight of the underlying evidence.

Wire provenance

This editorial synthesis draws on the following public wire/social posts:

• https://t.me/france24_en
• https://www.epa.gov/greenvehicles/fast-facts-us-greenhouse-gas-emissions