Microsoft and Meta slash Claude spending as Anthropic becomes rival

The timing is what matters here. Both pullbacks are happening simultaneously with Anthropic's IPO process underway, meaning public market investors will be pricing customer concentration risk against a revenue base that may already be contracting at its largest nodes before the S-1 even closes.

Anthropic's IPO filing from late September (covered across The Decoder, TechCrunch, and The Verge) showed 12x revenue growth alongside $8 billion in operating losses, with the prospectus itself flagging existential and business risks. What those filings could not fully capture was how quickly the enterprise customer base might shift. The $100 million engineer training commitment announced October 2nd reads differently now: if Meta and Microsoft are reducing Claude dependency, Anthropic's path to sustainable revenue runs through a much broader, more distributed customer base, and that bet only pays off if the practitioner ecosystem it's building actually generates sticky demand elsewhere. The civilian government pivot covered October 1st offers a partial hedge, but federal procurement cycles are slow and the Pentagon channel remains blocked.

Watch Anthropic's next disclosed revenue figures after IPO for whether enterprise API revenue concentration has shifted toward government and mid-market accounts. If the top-ten customer revenue share drops meaningfully from what the S-1 implies, the diversification strategy is working; if it holds or worsens, the structural vulnerability this story identifies will be a recurring line item in quarterly investor calls.