The debate on the impact of artificial intelligence on employment turned sharply more complex on June 29, 2026. A Ramp and Revelio Labs report covering 22,000 companies shows entry-level roles up 12 percent at heavy AI adopters. Goldman Sachs, on the opposite side, estimates 16,000 monthly jobs erased by AI over the past year.
Key Takeaways
- The Ramp Revelio Labs report shows entry-level jobs up 12 percent at high-intensity AI adopters
- Goldman Sachs estimates roughly 16,000 jobs per month lost to AI over the past twelve months
- 90,000 layoffs officially tied to AI have been counted from January through May 2026 in the United States
Have an AI Sum Up This Article
ChatGPTTwo Opposite Narratives Over the Same Window
On June 29, 2026, a wrap-up piece consolidates the recent publications on employment and artificial intelligence. The central finding is simple: available data tells two different stories over the same window. On one side, heavy AI adopters hire more. On the other, aggregate statistics show net losses that keep accelerating.
The Ramp and Revelio Labs report covers roughly 22,000 companies. The most-quoted result is the 10.2 percent headcount increase at high-intensity AI adopters, with a rise of 12 percent on entry-level roles. That figure challenges the widely spread idea that AI first destroys junior hires.
The counter narrative comes from Goldman Sachs, which estimates AI erased about 16,000 jobs per month over the past twelve months. Compounded over the year, the bank lands at magnitudes above 190,000 evaporated positions. The method differs from Ramp, but both figures coexist in the media and regulatory landscape.
The official count of layoffs attributed to AI in the United States reached 90,000 positions from January through May 2026. That number, built from public company announcements, does not measure the net effect but the frequency with which AI is cited in official reasons. The methodological gap explains part of the Ramp Goldman divergence.
The Boston Consulting Group adds a five-year projection: 15 percent of US jobs would be exposed to direct AI replacement by 2031. That projection, more forward-looking than observational, sits alongside Goldman’s monthly flow figures and Ramp’s headcount stocks.
The Methodological Bias That Reshuffles the Deck
The Ramp report explicitly acknowledges its limits. The authors state that their study “does not show that AI universally creates jobs”, but that it counters claims that AI would lead to broad job losses. That nuance matters and is often lost in the public reuse of the 12 percent figure.
The main identified bias is the sample. The 22,000 companies analyzed lean toward tech-forward, well-funded firms, already in a growth phase. In other words, we are observing companies for which AI adoption and growth are perhaps two effects of the same cause, rather than AI driving headcount growth.
The authors go further. They warn that resource-rich firms could widen their advantage over players experimenting with AI through simple subscriptions. In other words, the treatment gap between top adopters and the rest of the market may be the real takeaway of the report, more than the aggregate figure itself.
This reading reframes the question. The relevant debate is no longer “does AI create or destroy jobs”, but “which companies hire, which lay off, and why”. That reframing shifts the analysis toward inter-company inequality rather than a homogeneous macro effect.
This methodological divergence compounds with signals already covered in recent weeks. Anthropic announced the end of its junior hires at the entry-role level, even as the Ramp report counts entry-level gains at heavy adopters. A concrete example on the industrial side runs the same direction: Ford rehired 350 engineers after its AI quality push failed. Both realities coexist because they describe different market segments.
Also on Horizon:
- Claude Sonnet 5 Cheaper Than GPT-5.5 for AI Agents
- DSpark: DeepSeek Boosts AI Speed 85% on GPUs
- Meta Bans Claude Code and Codex for Engineers
What the Debate Reveals About the AI Jobs Link
Reading the three figure sets together produces a finer picture than any of them alone. Heavy AI adopters keep hiring, but probably because they already had the resources to invest simultaneously in the technology and the talent. The segmented market between them and the rest of the corporate base hardens further.
The gap with Goldman’s 16,000 monthly losses then reflects less a contradiction than a different focus. Goldman looks at the net macro flow, Ramp looks at the stock inside a specific corporate subset. Adding the two would be technically incorrect, but opposing them in the press is equally misleading.
One data point reinforces this reading. The Anthropic survey showing Claude handles half the tasks describes a shift at the individual level, not necessarily at the position level. In other words, an employee may see their workload change without losing their job, which makes net job counting a poor tool for fine-grained AI impact analysis.
In the short term, the consequence for policymakers is concrete. The BCG projection of 15 percent of jobs threatened by 2031 will feed regulatory arguments. Union representatives, especially on the US and European public sides, now hold a figure impactful enough to weigh in collective bargaining negotiations.
In the medium term, the dominant question concerns employees at mid-sized firms, neither tech-forward nor heavy adopters. These workers land in a gray zone. Their employer does not invest enough to benefit from Ramp-style effects, but likely uses AI tools enough to be exposed to the Goldman-style risk. That intermediate category is where the public debate will focus in the coming months.
Follow the story on Horizon.
