OpenAI Group PBC is an American artificial intelligence research organization and [[OpenAI Foundation|the world's most prominent AI company]], headquartered in an unmarked office building in San Francisco's Mission Bay neighborhood at 1515 Third Street, subleased from Uber. Founded in December 2015 as a nonprofit by [[Elon Musk]], [[Sam Altman]], [[Ilya Sutskever]], [[Greg Brockman]], and others, OpenAI triggered the global generative AI boom with the launch of [[ChatGPT]] in November 2022. By mid-2026, the company had grown to over 900 million weekly active users across its products and reached a valuation of $852 billion, while simultaneously facing the most intense competitive and regulatory pressure in its history.
In June 2026, OpenAI filed a confidential Form S-1 with the SEC for an anticipated initial public offering, joining [[Anthropic]] and SpaceX in a multi-trillion-dollar AI IPO pipeline. By late June, however, the company was reportedly leaning toward delaying the listing to 2027 as Altman refused to lower a $1 trillion valuation target, calling any reduction a "nonstarter," amid a tech stock slide following SpaceX's IPO and plateauing user growth near 900 million weekly active users. Leaked 2025 financial statements showed revenue of $13.07 billion but an operating loss of $20.92 billion, with total expenditure of $34 billion. In the first quarter of 2026, the company generated $5.7 billion in revenue but reported a GAAP net loss exceeding $21.3 billion — including $12.4 billion in non-cash restructuring charges — with $3.7 billion in cash burn, $8.6 billion in R&D spending, and off-balance-sheet computing procurement commitments estimated at $665 billion through 2030. OpenAI held over $73 billion in cash and marketable securities at quarter-end. The company planned to grow headcount from approximately 4,500 to 8,000 by year-end, and was considering substantial token price cuts in response to Anthropic's surging enterprise growth. Altman stated in a December 2025 podcast that he was "0 percent" excited about being a public company CEO but acknowledged the need for public market participation in value creation.
GPT-5.6 (Sol/Terra/Luna) was unveiled on June 26, 2026 in a limited preview — the first instance of a U.S. government requiring a frontier AI company to restrict access to a new model for national security review before public availability. The three-tier model suite featured Sol (flagship, $5/$30 per million tokens, nearly half the cost of Anthropic's Claude Fable 5), Terra (mid-tier at half Sol's price, similar in performance to GPT-5.5), and Luna (lightweight at half Terra's price). Sol introduced Max Reasoning Effort and Ultra Mode with coordinated sub-agents, and will be available on [[Cerebras Systems]] hardware in July 2026 at up to 750 tokens per second. The Trump administration approved approximately 20 partners on a case-by-case basis, following a June 2 executive order establishing voluntary pre-release AI review and an earlier directive forcing Anthropic to suspend its Mythos 5 and Fable 5 models. OpenAI stated the delay was temporary and opposed the process becoming the "long-term default."
On June 24, 2026, OpenAI also released a major update to GPT-5.5 Instant, the default model in free ChatGPT, improving intent recognition, multi-constraint reasoning, shopping recommendations, and conversational tone — described by the company as "much more fun to talk to." On June 17, OpenAI launched a dedicated Scheduled Tasks hub in ChatGPT, giving users centralized management of automated tasks and signaling the sunset of the Pulse daily summary feature.
The company began the most significant transformation of ChatGPT since its launch, turning it into a "superapp" that integrates coding tools, AI agents, and third-party services with [[Codex]] at the center. Thibault Sottiaux, promoted to head core product and platform, described the vision as a "personal AGI" — a unified interface that grows more capable and personalized over time. The product consolidation, formalized in a May 2026 reorganization under Greg Brockman, merged ChatGPT, Codex, and the API into a single core product team, with Nick Turley moving to lead enterprise industries and Ashley Alexander taking over consumer products. Fidji Simo remained on medical leave during the reorganization with expectations of returning. Visa embedded its payment network inside ChatGPT in June 2026, enabling AI agents to independently shop and complete transactions with spending limits and approval guardrails. OpenAI also launched its advertising business globally, activating Japan and South Korea as the first non-English markets while expanding the self-service Ads Manager to UK advertisers. At Cannes Lions in June 2026, CRO Denise Dresser introduced the "Intelligence Economy" concept, framing ChatGPT advertising as a utility-driven alternative to traditional search ads, with a target of $100 billion in ad revenue by 2030.
GPT-5.5 (Spud), the first fully retrained base model since GPT-4.5, was released on April 23, 2026, featuring a 1-million-token context window and pricing of $5/$30 per million tokens. The model achieved 82.7 percent on Terminal-Bench 2.0 and 85.0 percent on ARC-AGI-2, and was classified as "High" risk under OpenAI's Preparedness Framework for both biological/chemical and cybersecurity capabilities.[^c60] A specialized GPT-5.5-Cyber variant was released to vetted defenders in May 2026. On June 24, OpenAI unveiled Jalapeño, its first custom AI inference ASIC co-designed with Broadcom in nine months, matching Nvidia Blackwell performance with approximately 50 percent cost improvement, with initial deployment targeted by end of 2026. The Five Eyes intelligence alliance issued a rare joint warning in June 2026 that frontier AI models are "months, not years" away from transforming cyber capabilities, explicitly referencing OpenAI's Daybreak program.
OpenAI significantly expanded its Daybreak cybersecurity initiative on June 22, 2026, launching the Patch the Planet open-source security program in collaboration with Trail of Bits and HackerOne, an updated GPT-5.5-Cyber model scoring 85.6 percent on CyberGym, the Codex Security plugin that has scanned over 30 million commits across 30,000 codebases, and the Daybreak Cyber Partner Program with Accenture, Cisco, CrowdStrike, Cloudflare, and other major security vendors. The company announced one of its largest enterprise deployments to date — Samsung Electronics rolling out ChatGPT Enterprise and Codex to more than 120,000 employees across South Korea and its global Device Experience division — and signed a strategic partnership with Visa to embed payment infrastructure inside ChatGPT for AI-agent commerce. Samsung's deployment built on an earlier $71 billion high-bandwidth memory supply agreement signed in October 2025 for the Stargate project.
OpenAI announced that ChatGPT would debut on the Pentagon's GenAI.mil platform in early July 2026, serving over three million defense personnel. The Pentagon required an "all lawful uses" clause preventing company-imposed restrictions on military applications — a condition that OpenAI accepted after months of internal deliberation, and which rival Anthropic rejected, leaving Claude absent from the platform. The deal sparked employee backlash, with the company's robotics head resigning in protest. Altman later told staff the company was exploring a contract to deploy on all NATO classified networks.
The company deepened its presence in South Korea as Sam Altman met with Samsung, Kakao, and Naver in June 2026, while OpenAI recruited [[Noam Shazeer]] — the Transformer co-author and former Google DeepMind Gemini co-lead — as head of architecture research and [[Dean Ball]], former White House AI policy official, to lead a new Strategic Futures policy team. OpenAI also acquired Ona, a secure cloud execution startup, to strengthen Codex with persistent agent environments, and launched enterprise spend controls for ChatGPT Enterprise to help organizations manage AI costs. The "ChatGPT for Kakao" service surpassed 11 million cumulative users, and Kakao launched Kanana 2.5, its 150-billion-parameter agentic AI model. On June 27, OpenAI appointed Prabhjeet Singh, former Uber India head, as its Managing Director for India with a mandate spanning consumer growth, enterprise adoption, and partnerships.
OpenAI entered advanced negotiations to lease a 10-gigawatt AI data center campus in Ohio from SoftBank's SB Energy on former Cold War-era federal land, with NVIDIA providing credit guarantees for what would become the largest known AI infrastructure project. The U.S. Department of Energy confirmed a public-private partnership in March 2026 backing the project with $33.3 billion in Japanese funding. SB Energy confidentially filed for a US IPO in May 2026. In a related development, OpenAI's chip partner [[Cerebras Systems]] went public on Nasdaq at a $185 per share IPO price, raising $5.55 billion in the largest tech debut of 2026. The OpenAI-Cerebras relationship, valued at over $20 billion, became the subject of a California Attorney General investigation request after sworn testimony in the Musk trial revealed that Greg Brockman had held undisclosed personal Cerebras shares since 2017 while lobbying for OpenAI business with the chipmaker.
Amazon MGM Studios declined to release Artificial, a completed $40 million Luca Guadagnino biographical film about Sam Altman, citing a conflict of interest with Amazon's $50 billion strategic partnership with OpenAI. Getty Images signed a licensing deal to bring its stock photography library into ChatGPT search, sending Getty shares surging as much as 145 percent and rescuing the company from a potential NYSE delisting. OpenAI expanded internationally through its first Brazilian media partnership with Folha de S.Paulo and UOL, continued its collaboration with the IndiaAI Mission focused on voice-first AI, and partnered with JioHotstar for ChatGPT-powered streaming discovery.
The company faced escalating legal and regulatory headwinds. A multistate coalition of 42 attorneys general subpoenaed OpenAI on June 12, demanding internal documents on user data handling, minors' safety, advertising practices, and model sycophancy — the most significant regulatory escalation the company faced as it prepared for its IPO. A separate pre-litigation dispute emerged with Apple over the ChatGPT-Siri integration that fell far short of revenue projections, with OpenAI retaining external counsel for potential breach-of-contract action. OpenAI disclosed and disrupted two China-linked influence operations — "Data Center Bandwagon" and "Tech and Tariffs" — that used ChatGPT to generate content targeting U.S. data center and tariff debates. Florida Attorney General James Uthmeier filed the first state-led lawsuit against OpenAI and Sam Altman on June 1, alleging 10 counts including deceptive trade practices, negligence, and strict product liability, and a separate Canadian lawsuit alleged ChatGPT encouraged a 24-year-old woman to die by suicide, adding to 19 similar consolidated cases in California state court. In a separate controversy, CFO Sarah Friar suggested at a WSJ Tech Live event that the U.S. government should backstop AI infrastructure loans, prompting accusations of seeking taxpayer bailouts and forcing Altman to issue a thousand-word clarification rejecting government guarantees.
OpenAI's competitive position faced unprecedented pressure from Anthropic, which surpassed it in valuation for the first time ($965 billion versus $852 billion) and overtook it in enterprise AI adoption. Anthropic's annualized revenue run rate of approximately $45 billion exceeded OpenAI's roughly $25 billion, driven substantially by Claude Code, which propelled the company toward its first profitable quarter. At the G7 summit in Évian-les-Bains, Altman joined Anthropic CEO [[Dario Amodei]] and other AI leaders for a working lunch with world leaders, where Altman called for an international forum for AI safety standards and urged democratic governments not to leave regulation to companies alone. Anthropic also launched The Anthropic Institute, a dedicated research unit consolidating societal-impact and safety research under co-founder Jack Clark.
International expansion continued through multiple channels. SoftBank launched a cybersecurity patching service for Japan's critical infrastructure using OpenAI technology, with plans to increase cybersecurity staff twentyfold to 1,000. In South Korea, the "ChatGPT for Kakao" service surpassed 11 million cumulative users, and Kakao launched Kanana 2.5, its 150-billion-parameter agentic AI model. Brazil's first media partnership with Grupo Folha and Grupo UOL brought Brazilian journalism to ChatGPT. Armenia joined the Education for Countries initiative, and the University of Maine System awarded a $1.4 million ChatGPT Edu contract. GPT-5.5-powered AI agents reached a stage where they can operate autonomously for 24 hours without human intervention, up from 7 hours with GPT-5.
OpenAI's corporate transformation included converting to a public benefit corporation in October 2025, closing a $122 billion funding round at an $852 billion valuation in March 2026, and restructuring its product organization to merge ChatGPT, Codex, and developer APIs into a unified platform. In April 2026, OpenAI published an "Industrial Policy for the Intelligence Age" proposing a Public Wealth Fund to distribute AI-driven economic gains to all citizens, and engaged in discussions with the Trump administration about donating equity to seed such a fund. The company released GPT-5.5 (Spud), its first fully retrained base model since GPT-4.5, featuring a unified multimodal architecture and dual-system reasoning, and announced the retirement of GPT-4.5 and the o3 reasoning model from ChatGPT.