Source: Newcomer
A leaked investor presentation reveals the brutal unit economics required to justify Anthropic’s trajectory—the firm projected $14B in annual EBITDA losses on $18B revenue in 2026, suggesting the AI safety company will need to sustain massive cash burn to train increasingly capable models before achieving profitability. The $1.995T 2030 valuation target (a suspiciously precise miss of $2T) exposes how dependent frontier AI valuations are on faith in future breakthroughs rather than near-term business fundamentals, creating pressure on Anthropic to either dramatically outperform these assumptions or face a reckoning on unit economics that most software companies would never tolerate. Capital intensity and long development timelines have become the competitive advantage in AI—execution risk is massive, but so is the winner-take-most upside if scale effects eventually reverse the losses.