Executive Summary
Google (Gemini), OpenAI, and Anthropic are burning cash at vastly different rates, with divergent paths to profitability driven by business model, infrastructure ownership, and strategic focus.
Cash Burn & Profitability Timelines
| Company | 2025 Cash Burn | Revenue (2025) | Burn Rate | Profitability Target |
|---|---|---|---|---|
| OpenAI | ~$8–9B | ~$13–20B ARR | 70% of revenue | 2029–2030 |
| Anthropic | ~$3B | ~$5–9B ARR | 70% → 9% by 2027 | 2028 |
| Google (Gemini) | N/A (integrated) | Already profitable | N/A | Already profitable |
How Each Is Burning Cash
OpenAI: Growth-at-All-Costs
- Burn drivers: Massive compute costs ($8.67B in 9mo 2025), R&D ($6.7B in H1 2025), stock compensation ($2.5B in H1 2025), and infrastructure deals ($1.4T committed over 8 years).
- Why so high? OpenAI is betting on AGI/superintelligence, diversifying into video (Sora), hardware, e-commerce, and ads—all capital-intensive. It relies on Microsoft Azure for compute, paying the "NVIDIA tax" (GPUs cost $20K–35K vs. $3K–5K manufacturing cost).
- Cumulative losses: Projected $115B–143B by 2029.
- Revenue model: 55–60% consumer subscriptions (ChatGPT Plus/Pro), 25–30% enterprise, 15–20% API. Revenue doubled in H1 2025 to $12B ARR, but losses are growing faster.
Anthropic: Enterprise-First Discipline
- Burn drivers: Model training, multi-cloud infrastructure (AWS/Google/Azure), and R&D—but more capital-efficient than OpenAI ($2.10 revenue per dollar of compute vs. OpenAI's $1.60).
- Why lower? Anthropic focuses on B2B (80% of revenue), avoids compute-heavy image/video generation, and targets higher-margin enterprise contracts. Claude Code is a breakout success (~$1B run-rate).
- Path to profit: Cash burn drops to 33% of revenue in 2026, 9% in 2027, and breakeven by 2028.
- Revenue model: Enterprise API (85% of revenue), 300K+ business customers, $5–9B ARR in 2025.
Google (Gemini): Already Profitable
- Burn drivers: Google is spending $180B+ in 2026 CapEx (largest in tech history) on AI infrastructure, but this is self-funded through Alphabet's $73B+ annual free cash flow.
- Why profitable? Gemini is integrated into Google's existing products (Search, Workspace, Android), so it monetizes through ads, cloud, and subscriptions without needing standalone profitability. Google owns its full stack: custom TPUs (6x cheaper than NVIDIA GPUs), data centers, and 3B+ Android devices for distribution.
- Revenue model: Google Cloud hit $70B run-rate in Q4 2025 (30% operating margin). Gemini drives incremental revenue across Search, Workspace, and GCP—no separate P&L needed.
Paths to Profitability
OpenAI: Scale or Die
- Strategy: Achieve AGI/superintelligence to unlock massive efficiency gains and new markets (labor replacement, enterprise automation). Monetize via ads (launching 2026), enterprise seats (3M paying business users), and API licensing.
- Risk: Profitability hinges on $200B revenue by 2030 and 60% margins—analysts call this "dotcom-era extrapolation." If revenue growth slows or compute costs spike, the model collapses.
- Dependency: Microsoft owns 27% of OpenAI and provides Azure compute. Full acquisition is plausible if cash burn accelerates.
Anthropic: Lean & Enterprise-Focused
- Strategy: Target regulated industries (finance, healthcare) with "Constitutional AI" (safety/compliance focus). Expand enterprise API revenue and maintain capital efficiency.
- Risk: Smaller user base (18.9M MAUs vs. ChatGPT's 350–450M) limits consumer upside. Must compete with Google's ecosystem lock-in and OpenAI's brand.
- Advantage: Clearest path to profitability—positive free cash flow by 2027, breakeven by 2028.
Google: Ecosystem Dominance
- Strategy: Integrate Gemini into Search, Workspace, and Android to protect core ad revenue and expand cloud market share. Use TPU cost advantage to undercut rivals on pricing.
- Risk: Gemini's independent market share (13–15%) lags ChatGPT (60%). Must prove Gemini drives incremental revenue, not just cannibalizes Search.
- Advantage: Already profitable. Can subsidize AI losses indefinitely with $70B+ annual free cash flow.
Key Takeaways
- OpenAI is burning the most cash ($8–9B/year) to chase AGI and diversify revenue streams, but profitability is 4–5 years away and depends on heroic assumptions.
- Anthropic is burning less ($3B/year) and targeting profitability by 2028 through enterprise focus and capital discipline.
- Google is already profitable, spending heavily on infrastructure but monetizing AI through its existing ecosystem (Search, Cloud, Workspace).
The winner isn't who burns the most—it's who converts cash into defensible revenue streams. OpenAI bets on AGI breakthroughs, Anthropic bets on enterprise trust, and Google bets on ecosystem lock-in.