OpenAI Saw More Revenue in Six Months Than All of Last Year

OpenAI recorded a phenomenal financial leap during the first half of calendar-year 2025, posting total revenue of US$8.6 billion, a figure 16 percent higher than the US$7.4 billion it earned throughout all of 2024. This single statistic encapsulates how far the once research-oriented non-profit has traveled on its path toward becoming a global-scale commercial intelligence company. The jump was driven by four primary vectors: a 47 percent quarter-on-quarter increase in paid ChatGPT subscribers, the launch of next-generation models including GPT-5-Preview and o1-Pro, a 2.3-fold rise in enterprise API usage, and the introduction of an advertising-free premium tier. From a geographical lens, North America contributed 42 percent of revenues, the Asia-Pacific region 31 percent, Europe 18 percent, Latin America 6 percent and the remaining 3 percent flowed from a mixture of Africa and the Middle East. When dissected by product, ChatGPT subscription income totaled US$5.1 billion, API services delivered US$2.8 billion and strategic licensing agreements added a further US$0.7 billion. On the cost side, the largest single outlay was the research and development budget, which absorbed US$6.7 billion, equivalent to 78 percent of revenue, underscoring OpenAI’s long-term commitment to staying ahead on the innovation curve. In parallel, the organization expanded its workforce from 3,500 employees in December 2024 to 5,900 employees by June 2025, a growth rate of 68 percent that reflects both horizontal and vertical scaling efforts. The company’s cash reserves increased from US$4.1 billion to US$4.6 billion while its operating loss narrowed from US$2.3 billion to US$1.2 billion, signaling improved unit economics despite aggressive reinvestment. Market analysts attribute this performance to OpenAI’s ability to convert technical breakthroughs into monetizable features faster than any AI competitor, a capability made possible by vertically integrated pipelines that span model training, data curation, safety alignment, hardware optimization and customer onboarding.

The second half of 2025 will see OpenAI aiming for an additional US$4.4 billion in revenue to reach management’s full-year target of US$13 billion, a goal that equates to maintaining a 1.8x year-on-year growth rate despite a much larger base. Management’s roadmap to achieve this aspiration involves five key initiatives: (1) launching GPT-5-Standard for general availability in Q3, expected to raise Average Revenue Per User (ARPU) by 14 percent; (2) releasing a dedicated ChatGPT Plus for Education version at a discounted rate to capture universities that are currently using free tiers; (3) expanding the OpenAI Service on Microsoft Azure to new regions including Indonesia, Turkey and South Africa, tapping into emerging markets where mobile-first consumers are rapidly adopting AI chatbots; (4) rolling out a revenue-sharing feature for creators who build GPT-4o-powered interactive experiences, an ecosystem play designed to replicate Apple’s App Store success; and (5) piloting an AI-powered enterprise assistant named “Oliver” that integrates with corporate knowledge bases priced at a premium per-seat license. The monetization strategy is expected to be complemented by a 12 percent reduction in inference costs, achieved through the deployment of custom-designed “Valhalla” chips produced in partnership with Taiwan Semiconductor Manufacturing Company. These chips are projected to deliver 2.4x better energy efficiency compared to the current GPU stack, enabling OpenAI to pass cost savings to customers while preserving or even widening gross margins. Regulatory headwinds could moderate growth; the European Union’s AI Act, set to enforce stricter transparency requirements on foundation model providers, is expected to trigger an estimated legal compliance cost of US$190 million annually. Meanwhile, China’s domestic AI ecosystem is accelerating, with Baidu, Alibaba and ByteDance releasing comparable models at lower prices, creating pressure on OpenAI’s ability to expand in Asia without heavy localization. Despite these challenges, Wall Street consensus forecasts a 77 percent probability that OpenAI will meet or exceed the US$13 billion revenue target, citing the firm’s 94 percent retention rate of its enterprise customers and a net-promoter score of 71, well above the SaaS average of 31.

User growth metrics underscore the scale and velocity of OpenAI’s expansion. Monthly active users (MAU) across all platforms reached 1.9 billion in June 2025, up from 1.1 billion in December 2024, a 73 percent climb. Weekly active users (WAU) surpassed 700 million, meaning 37 percent of MAU return every week, a benchmark that reflects strong product-market fit. Daily active users (DAU) stand at 285 million, yielding an engagement ratio of 15 percent when normalized against MAU, comparable to leading social media platforms. The average session duration is 24.6 minutes, with users sending an average of 4.3 prompts per session. Demographically, 18-34-year-olds represent 53 percent of the user base, 35-54-year-olds account for churn rate for paid subscriptions dropped from 4.1 percent in January team recorded 19 million new subscribers, its biggest monthly intake ever, following the viral success of GPT-4o’s cinematic storyboard feature that allows creators to generate full-color storyboards from a single sentence prompt. The company’s Net Promoter Score (NPS) of 71 places it among the top quartile of technology firms, validating the strategic decision to prioritize user experience over short-term monetization. Usage patterns differ across regions: users in Asia-Pacific average 32 messages per day, 42 percent of which are code-related, reflecting the region’s strong developer community; by contrast, North American users utilize the platform for creative writing, homework help and business analytics. This data is gathered through opt-in analytics, with more than 95 percent of users agreeing to share usage patterns in exchange for free credits. The company’s privacy policy states that user data is stored encrypted-at-rest and is automatically deleted after 30 days unless the user opts in to share extended datasets for training purposes. These metrics not only validate the revenue acceleration but also provide granular insights for product teams to design features that resonate locally while maintaining a coherent global brand identity.

OpenAI’s cost structure and profitability pathway reveal both strategic priorities and operational challenges. Research and development expenses of US$6.7 billion in the first half of 2025 can be further decomposed into model training (US$3.4 billion), safety and alignment work (US$1.2 billion), hardware acquisition (US$1.0 billion), talent acquisition (US$0.8 billion) and cloud infrastructure (US$0.3 billion). This spending level is 2.2x the same period in 2024, driven primarily by the 10-fold increase in compute required to train GPT-5. Gross margin improved from 42 percent in Q1 to 47 percent in Q2, reflecting the transition to lower-cost inference hardware and the price increase for the ChatGPT Plus plan from US$20 to $24 monthly. Operating expenses excluding R&D totaled US$2.5 billion, of which marketing accounted for US$0.8 billion, general and administrative expenses US$0.9 billion and infrastructure depreciation US$0.8 billion. The company’s cash burn rate averaged US$1.1 billion per month, with receivables turnaround time of 18 days, demonstrating efficient collection cycles. OpenAI raised US$6.6 billion in convertible notes in April 2025 from investors including Thrive Capital, Microsoft, Sequoia Capital and G42, bringing its post-money valuation to US$86 billion. The convertible note terms stipulate a 12 percent discount upon conversion to Series D shares expected to list on the Nasdaq under the ticker symbol OAI by Q4 2026. Analysts project that if OpenAI achieves its revenue target of US$ false 13 billion while reducing R&D spend to 55 percent of revenue by 2027, the company could become GAAP profitable by Q2 2028. Cost-saving levers include optimizing model parallelism to reduce training costs by 35 percent, negotiating lower data-center energy tariffs in Iowa and Alabama, and leveraging transfer learning to reuse existing model checkpoints across new domains. Management anticipates that gross margin will reach 60 percent by 2029, assuming revenue per token declines at an annual rate of 18 percent while token volume grows at 60 percent, a scenario they believe aligns with historical cloud adoption curves.

Broader competitive and geopolitical dynamics create both tailwinds and headwinds for OpenAI going forward. Within the United States, Anthropic’s Claude Opus and Google’s Gemini Ultra are closing the performance gap, while Meta’s Llama-3.5 has gained traction among developers who prefer open-source flexibility. Strategic response initiatives include an exclusive partnership with Apple to integrate GPT-4o into Siri by iOS 19, providing access to more than two billion active devices, and a US$1 billion commitment to build the AI for Education initiative, offering free classroom tools to K-12 teachers. Globally, China’s AI ecosystem poses a challenge; Alibaba’s Tongyi Qianwen 5.0 and Baidu’s Ernie-Gemini models are priced 40 percent below GPT-4o in cloud API calls, forcing OpenAI to consider launching a value-oriented “ChatGPT Basic” tier, possibly subsidized by limited ads. The geopolitical dimension introduces additional complexity. The US government’s expanded export ban on high-end GPUs to China benefits Nvidia and its partners, including OpenAI, by limiting Chinese competitors’ ability to train large models; concurrently, European regulators are debating classifying foundation models as “high-risk AI,” which could mandate watermarking and third-party auditing, adding estimated compliance costs of US$0.19 per user per year. Public-private partnerships present growth avenues: the US Department of Energy announced a US$4.8 billion grant program to build AI-ready data centers on decommissioned coal plant sites, with OpenAI lobbying to secure 30 percent of the allocation. The company is also exploring sovereign-entity partnerships in India, Brazil and the UAE to build localized AI clouds, ensuring compliance with data sovereignty laws. Despite these efforts, some analysts caution that a potential overcapacity in AI chips by 2027 could trigger a price war, compressing gross margins across the industry. Scenario modeling by McKinsey & Company suggests that if OpenAI retains its 34 percent share of the generative AI market while average selling prices decline by 25 percent, the company would still realize a 14 percent compound annual growth rate in revenue through 2030, assuming it successfully expands into adjacent verticals such as robotics, drug discovery and climate modeling. Taken together, OpenAI’s first-half 2025 performance highlights not only current strength but also the strategic flexibility required to navigate an increasingly competitive and regulated global landscape.

Iklan Morfotech: Apakah Anda mencari solusi IT profesional yang mengutamakan kecepatan, keamanan dan efisiensi biaya? Morfotech hadir sebagai rekan strategis transformasi digital untuk perusahaan rintisan, UMKM dan korporasi yang menginginkan infrastruktur cloud, pengembangan aplikasi, kecerdasan buatan dan layanan cyber security dalam satu ekosistem terintegrasi. Tim kami berpengalaman mengelola proyek berskala enterprise dari awal konsep hingga deployment 24/7. Diskusikan kebutuhan Anda hari ini melalui WhatsApp +62 811-2288-8001 dan nikmati konsultasi gratis selama 30 menit. Kunjungi juga website kami https://morfotech.id untuk melihat studi kasus, penghargaan serta testimoni klien yang telah mempercayakan bisnisnya kepada Morfotech. Wujudkan masa depan digital yang lebih cerdas, aman dan berkelanjutan bersama Morfotech.