Inside OpenAI's Billion-User Milestone: The 80% Luna Price Cut, the $38.5B Loss, and the AI Price War

Why the Cuts Happened: A Three-Way Competitive Squeeze

The cuts were a direct response to rivals offering comparable or superior performance at a fraction of OpenAI's original pricing.

Anthropic Claude Sonnet 4.6 (released February 2026) delivers near-flagship coding and agentic performance at $3/$15 per million tokens, with an introductory discount to $2/$10 through August 31, 2026 . It scores 79.6% on SWE-bench Verified and led all models in office productivity benchmarks .

DeepSeek V4 Flash (released April 2026) is far more aggressive: $0.14 input / $0.28 output per million tokens—roughly 1/50th of Luna's original price and 1/100th of Sonnet 4.6's—yet competes strongly on reasoning and agentic tasks . One analysis pegs DeepSeek V4 Flash at an effective $0.06 per million tokens versus Claude's $2.31 . DeepSeek put V4 Flash API into public beta on July 31, 2026, further intensifying the competition .

Against this landscape, OpenAI's original Luna pricing ($1/$6) was uncompetitive for high-volume API workloads. The 80% Luna cut brings it closer to cost parity with Anthropic's introductory pricing, though DeepSeek remains dramatically cheaper.

What the Financials Reveal: $13 Billion Revenue, $34 Billion Burn

OpenAI's audited 2025 financials—leaked June 2026 by independent journalist Ed Zitron and independently verified by the Financial Times—paint a stark picture :

Key takeaways from the documents:

• Revenue more than tripled year-over-year, exceeding the company's own $10B internal target, and annualized revenue hit roughly $25B by early 2026 .
• Research and development alone ($19.18B) exceeded all revenue, with Microsoft Azure compute representing a massive cost line (estimated at $17B+) .
• The $38.5B net loss is inflated by a one-time $41.55B non-cash charge from OpenAI's October 2025 public-benefit corporation conversion and equity liability revaluation—the real operating burn was roughly $21B . The Financial Times put the adjusted loss closer to $8 billion .
• Deutsche Bank projects cumulative negative free cash flow of $143 billion through 2029 if the current trajectory holds .

What This Tells Us About the AI Industry Price War

Four patterns emerge from OpenAI's moves and financial position:

Market share is being bought, not earned

The 80% Luna cut three weeks after launch signals that initial pricing was set too high for the competitive reality—and that OpenAI is willing to sacrifice near-term margin to defend its developer ecosystem against Anthropic and DeepSeek.

Enterprise spend is bifurcated

OpenAI's 2 million business customers and 92% Fortune 500 penetration suggest enterprises pay a premium for reliability, brand, and ecosystem (Codex, ChatGPT Work, enterprise compliance) . But the API price cuts target the high-volume, price-sensitive developer segment where DeepSeek's 50-100x cost advantage is hardest to ignore.

The industry is in a 'build at any cost' investment phase

OpenAI's $34 billion in 2025 spending—$19.18 billion of which was R&D—mirrors a pattern across AI: massive infrastructure buildout, negative unit economics, and dependence on capital markets. OpenAI closed a $122 billion round at an $852 billion valuation in March 2026 , and filed a confidential S-1 in June targeting a valuation as high as $1 trillion . But the leaked financials raise hard questions about when—or whether—these companies become self-sustaining. OpenAI has told investors it aims to become profitable by 2030 .

Commoditization pressure is accelerating

When DeepSeek V4 Flash delivers strong reasoning performance at $0.14/$0.28 per million tokens, the differentiation premium for frontier models compresses rapidly . OpenAI's response—cut prices, bundle into subscriptions, and push higher-volume usage—is the classic playbook of a market transitioning from scarcity to commodity.

The Bottom Line

OpenAI's twin announcements on July 30-31, 2026 tell a single story: the company is trading billions in short-term revenue for user acquisition and ecosystem lock-in, betting that scale and efficiency improvements will eventually close a gap that currently costs it $3.30 for every dollar earned . With cash reserves of roughly $25 billion and a path to an IPO, OpenAI can afford that bet for now. Whether the math works by 2030—the company's self-imposed profitability deadline—will define the next phase of the AI industry.