Snap Resolves the Attribution Question — But the Wave Forecast Is Still Unfinished
We need to be precise about what just happened: Snap's announcement — a named executive, a specific metric (65% of new code), a specific headcount number (1,000 roles), and a specific dollar figure ($500M in savings) tied directly to AI capability — meets the threshold of explicit public attribution. That part of our thesis has arrived. What we're still forecasting at 73% is something harder: whether this becomes a wave — multiple major companies in the same period making similarly explicit public attributions, rather than a single data point surrounded by companies that are still speaking in euphemisms.
Let's separate the data streams, because conflating them has been a problem in how we've framed this forecast. Snap is direct evidence. One Fortune 500-scale company has done what we said would be the key signal: a CFO-level attribution linking a specific AI capability metric to specific headcount reduction in a public earnings context. That event has occurred. If our forecast had been 'will any major tech company explicitly attribute layoffs to AI in 2026,' it would be resolved YES and we'd be writing a retrospective.
But the forecast is about a wave — and the rest of the current data is substantially weaker. Meta's 8,000 cuts are framed as 'AI-infrastructure-focused restructuring,' which is meaningfully different from Snap's direct attribution. 'We're restructuring toward AI' is not the same as 'AI is now doing work that previously required these headcounts.' Amdocs (2,900 cuts) and Salesforce (86 cuts) issued no explicit AI attribution statements at all — they appear in the 100,000+ figure precisely because they're happening during an AI investment cycle, not because they've made Snap-style disclosures. Mustafa Suleyman's walk-back — explicitly reframing 'jobs' as 'tasks' — is itself evidence that Microsoft is actively managing attribution language in the opposite direction from Snap.
Our 73% reflects this split reality. The number went from 72% to 73% — a modest move — because Snap proves the behavior is possible and provides a template others can follow. It does not move further because the wave criterion requires demonstrated replication, and we don't have that yet. We're watching for whether Snap's stock reaction (+11%) functions as the incentive signal we hypothesize: if two or three other CFOs make comparably explicit attributions in Q2/Q3 earnings cycles, we'd move to 82-85%. If Q3 earnings season passes with Snap remaining the only explicit attribution, we'd revisit the 73% seriously downward. The counterargument that 'companies avoid PR risk of attribution' remains live — Suleyman's task/job reframe is the clearest evidence that the incentive structure hasn't universally shifted.
What keeps this at 73% rather than higher is also something we're being honest about: the forecast may be partially defining its own difficulty. The reason companies speak in euphemisms isn't irrational — there are legal, HR, and reputational reasons to avoid saying 'AI replaced these people.' Snap may turn out to be an outlier in disclosure culture rather than a leading indicator. We're watching Q3 earnings season as the decisive test. Specifically: if three or more S&P 500 companies include Snap-style attribution metrics in earnings disclosures between July and October 2026, the wave forecast resolves YES. If we reach November with Snap still standing largely alone, we're wrong about the wave and the probability drops to the low 50s.