Google’s Gemini reports 350M monthly users vs OpenAI’s 600M, facing EU antitrust probes and $4B annual losses as AI infrastructure costs strain profitability.
New disclosures reveal Google’s Gemini AI serves 350M monthly active users (MAUs), significantly trailing OpenAI’s 600M ChatGPT users. The EU Commission confirmed antitrust investigations on June 25 into Gemini’s Android integration under Digital Markets Act rules. With Morgan Stanley estimating $0.12/user monthly costs for Gemini versus OpenAI’s $0.20, both firms pursue enterprise deals like Google’s $30B Merck cloud partnership to offset losses. Alphabet shares fell 2.3% this week as CFO Ruth Porat acknowledged ‘material’ AI spending impacts.
The MAU Gap and Regulatory Crosshairs
Google disclosed 350M Gemini users this week through Workspace and Android integrations, while OpenAI’s standalone ChatGPT maintains dominance with 600M MAUs. EU competition chief Margrethe Vestager stated June 25: ‘We’re scrutinizing whether Gemini’s bundling creates unfair advantages under DMA Article 6.’ This follows Germany’s antitrust office demanding Google disclose Gemini data flows last month.
Infrastructure Costs Strain Margins
Morgan Stanley’s June 27 analysis reveals Gemini operates at 40% lower costs than OpenAI ($0.12 vs $0.20/user) thanks to custom TPUs. However, Alphabet’s AI division reportedly burns $4B annually according to The Information. ‘You’re seeing classic hyperscaler economics,’ said analyst Brian Nowak, ‘where every 100M users require $500M in quarterly capex.’
Enterprise Deals Anchor Monetization
Both firms target corporations to offset losses: Google’s June 24 Merck deal includes Gemini APIs for drug discovery, while OpenAI expanded Walmart and HSBC contracts this week. ‘Enterprise contributes 72% of our AI revenue despite 15% of users,’ revealed OpenAI COO Brad Lightcap in a TechCrunch interview.
Investor Patience Tested
Alphabet shares dipped 2.3% on June 26 after Porat’s margin warnings. Bernstein’s Mark Shmulik notes: ‘Markets tolerated AWS’ losses for a decade – but AI’s ROI timeline is compressed.’ Roku’s CFO warned of ‘AI fatigue’ as S&P 500 firms cut generative AI budgets by 18% this quarter per IDC.
Historical Parallels: From Browser Wars to Cloud
The EU’s scrutiny echoes its 2004 Microsoft investigation over Media Player bundling, which resulted in €497M fines. Like AWS’ early unprofitability, analysts compare AI’s ‘scale-first’ approach to Amazon’s decade-long cloud investment. However, antitrust remedies could force API interoperability – a 2024 twist absent in prior tech battles.
The Path Ahead: Regulated Utilities or Proprietary Moats?
As the EU AI Act advances, draft provisions mandate model interoperability by 2026. Gartner’s Arun Chandrasekaran warns: ‘Forced API access could reduce differentiation, turning foundation models into commodities.’ This contrasts with Microsoft’s proprietary Windows-Office moat that defined the PC era, suggesting new competitive dynamics for the AI age.
