AI industry faces a reality check after a year of record spending

In 2025, the artificial intelligence sector experienced a “vibe check”—a shift in industry sentiment from extreme optimism to skepticism. The year began with massive capital flows into AI labs and startups:

• OpenAI raised $40 billion at a $300 billion valuation, and is reportedly in talks to raise $100 billion at an $830 billion valuation.
• Anthropic secured $16.5 billion, pushing its valuation to $183 billion.
• Elon Musk’s xAI raised at least $10 billion.
• Thinking Machine Labs, founded by former OpenAI chief technologist Mira Murati, secured a $2 billion seed round at a $12 billion valuation.
• Lovable raised a $200 million Series A, followed by $330 million at nearly $7 billion.
• AI recruiting startup Mercor raised $450 million at a $10 billion valuation.

To support these valuations, AI’s biggest players promised close to $1.3 trillion in future infrastructure spending. Much of this capital was deployed through “compute-tied circular deals,” where investment funding flows directly back into the investor’s own cloud or chip ecosystem. Meta spent nearly $15 billion to lock up Scale AI CEO Alexandr Wang, while projecting $72 billion in capital expenditures for 2025. In October, Alphabet announced plans to lift its compute spend in 2026 to $93 billion, following its $4.75 billion acquisition of Intersect. Infrastructure developments included:

• Stargate, a joint venture to build AI infrastructure in the U.S. for up to $500 billion.
• A planned $10 billion Oracle data-center deal tied to OpenAI, which faced complications when a financing partner withdrew.
• Calls from U.S. Senator Bernie Sanders to rein in data center expansion.

This expansion triggered increased scrutiny over safety, ethics, and funding. The industry faced reports of “AI psychosis”—a term describing chatbots reinforcing user delusions—alongside copyright disputes, such as Anthropic’s $1.5 billion settlement. Tensions also emerged within the labs. Anthropic CEO Dario Amodei confessed in a leaked memo that he was “not thrilled” about taking money from dictatorial Gulf states.

As the pace of model breakthroughs slowed, the focus shifted from raw capabilities to distribution and business models. To prove economic value, OpenAI reportedly considered charging up to $20,000 per month for specialized AI. Meanwhile, Perplexity paid $400 million to power search inside Snapchat, buying its way into existing user funnels as startups race to build sustainable business models.

Why it matters

The AI industry in 2025 shifted from a period of unchecked investment and hype to a “vibe check” characterized by increased scrutiny, infrastructure constraints, and a need to prove sustainable business models.