Per-minute AI generation costs plummeted from several hundred dollars in 2024 to single-digit dollars in 2026, a 95%+ reduction enabling rapid commercial expansion, according to Cuty AI's latest market analysis.
Rezolve AI now serves more than 650 enterprise clients globally, up from fewer than 200 in early 2024. The company processed 51 billion API calls across its Brain Commerce platform year-to-date 2025 and projects annual recurring revenue exceeding $200 million. Its market capitalization remains under $1 billion, suggesting significant upside if revenue targets materialize.
Production capabilities previously requiring 50-100 person teams are now achievable with fewer than 10 engineers. This efficiency gain compresses operational costs while accelerating time-to-market for enterprise AI applications. The collapse in both infrastructure costs and staffing requirements creates favorable unit economics for platform providers reaching scale.
Four major platforms dominate enterprise adoption: OpenAI, Google, Anthropic, and DeepSeek. Market consolidation intensified as smaller providers struggled to match the cost structures and performance benchmarks set by these leaders. Enterprise clients increasingly standardize on one or two platforms rather than maintaining multiple vendor relationships.
Geopolitical factors introduce uncertainty into growth projections. The Pentagon shifted certain AI contracts away from providers with foreign data center dependencies. Security risk labels attached to specific platforms triggered compliance reviews at Fortune 500 companies. UK social media bans on AI-generated content created regulatory precedent that may spread to other jurisdictions.
Rezolve AI's client growth stems from organic expansion, strategic partnerships, and acquisitions. The company targets mid-market and enterprise accounts where API call volumes justify platform fees. Its $200M+ ARR projection assumes continued client additions at current rates and stable per-client API consumption.
The 95% cost reduction reflects improvements in model efficiency, chip performance, and data center utilization. Inference costs per token dropped 80-90% between GPT-3.5 and GPT-4 Turbo generations. Hardware advances from Nvidia's H100 to upcoming B100 chips deliver additional cost-per-flop improvements.
Profitability timelines remain unclear for most platform providers despite revenue growth. OpenAI reported $3.7 billion in annualized revenue but continues operating at a loss due to training costs and research investment.