--- title: "Uber gave 30,000 workers an AI agent and kept the price a secret" description: "Uber handed 30,000 employees an OpenAI agent without disclosing the price, in the same year it burned its AI budget in four months and capped per-worker AI spend at $1,500 a month. The undisclosed contract is a reference deployment for OpenAI's reported $1.25 trillion pre-IPO raise." publisher: "The Inference" section: "Applications" published: 2026-10-23T14:28:25.795Z modified: 2026-10-23T14:28:25.795Z canonical: https://theinference.org/article/uber-gave-30-000-workers-an-ai-agent-and-kept-the-price-a-secret language: en keywords: "Uber, OpenAI, ChatGPT, AI Agents, Enterprise AI, AI Spend" --- # Uber gave 30,000 workers an AI agent and kept the price a secret > Uber handed 30,000 employees an OpenAI agent without disclosing the price, in the same year it burned its AI budget in four months and capped per-worker AI spend at $1,500 a month. The undisclosed contract is a reference deployment for OpenAI's reported $1.25 trillion pre-IPO raise. [Uber](https://theinference.org/markets/companies/uber) gave every one of its more than [30,000 employees an AI agent on October 22](https://openai.com/index/uber-chatgpt-work-rollout/), scaling a 1,000-person pilot of [OpenAI](https://theinference.org/markets/companies/openai)'s ChatGPT Work thirty-fold in under a year [1]. OpenAI calls it a scale of deployment neither competing product, [Anthropic](https://theinference.org/markets/companies/anthropic)'s Claude Cowork nor [Microsoft](https://theinference.org/markets/companies/microsoft)'s Copilot Cowork, has publicly claimed [3]. [Uber will not say what it pays for it](https://techcrunch.com/2026/10/22/uber-puts-all-30000-employees-on-chatgpt-work/) [2]. The missing price is the story. ChatGPT Work is not a chatbot: it breaks a goal into steps, runs them for hours across Gmail, Google Drive, Calendar and Slack, and returns finished work, with GPT-6.1 Sol as the default model [1][3]. An agent like that is a consumption product. Every hour every employee lets it run is metered spend. ## A consumption-priced agent on every desk meets a $1,500-a-month cap Set that against Uber's own behavior this year. It [exhausted its entire 2026 AI budget in four months](https://theinference.org/article/uber-s-53-price-climb-built-its-profits-ai-agents-may-claw-them-back), then [capped employees at $1,500 a month for each AI coding tool](https://www.thewealthadvisor.com/article/visionary-square-paying-its-advisors-ai-prospecting-finny-gets-250-more-doors-open) [4]. A company rationing AI per person, per tool, just handed every knowledge worker a system built to consume tokens continuously. Uber's VP of engineering, Sumanth Sukumar, says the company will track usage and evaluate outcomes before deciding how far AI goes next [2]. That is a buyer still measuring, not a buyer with a savings number to announce. ## The logo is worth more than the check OpenAI needs this reference customer. Its enterprise revenue now generates more revenue than consumer, [a crossover its CFO announced on August 14](https://www.techtimes.com/articles/329404/20261023/uber-puts-agentic-ai-every-desktop-chatgpt-work-rolls-out-all-30000-employees.htm); business revenue grew 32% in a single month in July, and the run rate sits near $40 billion [3]. The same week, it was reported to be [raising up to $30 billion at a $1.25 trillion pre-money valuation](https://techcrunch.com/2026/10/22/openai-working-on-new-30b-raise-at-1-25t-valuation-ahead-of-ipo/), ahead of confidential IPO paperwork filed in June [5]. A 30,000-seat deployment is the proof for that pitch [5]. A marquee deal with no findable price leaves the revenue-quality question open. *Chart: **OpenAI's reported marks climb toward the IPO** Column chart of OpenAI's reported private valuations, split by basis: $852 billion post-money from the March 2026 raise, and the reported September Bloomberg target of $1.4 trillion and October pre-money mark of $1.25 trillion, both pre-money.* *OpenAI's reported private marks, split by basis: post-money from a closed raise, pre-money for terms still being negotiated. Bloomberg's $1.4 trillion (September) and the $1.25 trillion pre-money (October) are reports, not closed deals. [5]* ## What the November print will show Inside Uber the picture is heavier. It cut about 3,300 jobs in September, roughly 10% of the workforce, and about 10% of customer service roles in July; the CEO stopped short of calling either an AI cut [3]. [Gartner](https://theinference.org/markets/companies/gartner) expects 40% of companies deploying autonomous agents to demote or decommission them after governance failures [3]. [Uber reports Q3 on November 4](https://theinference.org/article/uber-s-q3-report-must-reconcile-22-bookings-growth-with-12-revenue-growth), the first print where any of this cost could surface [3]. Watch whether the agent shows up as headcount falling or opex rising. ## Takeaway A 30,000-seat agent deal with no findable price is OpenAI's reference deployment for a reported $1.25 trillion pre-money raise; Uber's November 4 print shows whether it costs headcount or opex. ## Sources 1. [OpenAI, Uber brings ChatGPT Work to all employees with OpenAI, 22 October 2026](https://openai.com/index/uber-chatgpt-work-rollout/) 2. [TechCrunch, Uber puts all 30,000 employees on ChatGPT Work, 22 October 2026](https://techcrunch.com/2026/10/22/uber-puts-all-30000-employees-on-chatgpt-work/) 3. [Tech Times, Uber Puts Agentic AI on Every Desktop, 23 October 2026](https://www.techtimes.com/articles/329404/20261023/uber-puts-agentic-ai-every-desktop-chatgpt-work-rolls-out-all-30000-employees.htm) 4. [The WealthAdvisor, Corporate America Is Getting Tired Of Paying Top Dollar for AI, October 2026](https://www.thewealthadvisor.com/article/visionary-square-paying-its-advisors-ai-prospecting-finny-gets-250-more-doors-open) 5. [TechCrunch, OpenAI working on new $30B raise at $1.25T valuation ahead of IPO, 22 October 2026](https://techcrunch.com/2026/10/22/openai-working-on-new-30b-raise-at-1-25t-valuation-ahead-of-ipo/) --- Uber gave 30,000 workers an AI agent and kept the price a secret — The Inference. Canonical: https://theinference.org/article/uber-gave-30-000-workers-an-ai-agent-and-kept-the-price-a-secret