Meta Platforms, Inc. vs OpenAI: Strategic Comparison

This in-depth comparison examines Meta Platforms, Inc. and OpenAI across revenue, market value, business model, competitive positioning, and long-term growth strategy. Whether you are researching Meta Platforms, Inc. on its own, evaluating OpenAI, or weighing the two companies side by side, the breakdown below highlights where each company leads and where the gap between Meta Platforms, Inc. and OpenAI is widest.

On the headline numbers, Meta Platforms, Inc. reports annual revenue of $201.0B against $5.0B for OpenAI, while their respective market capitalizations stand at $1.55T and $300.0B. Meta Platforms, Inc. is headquartered in United States and OpenAI operates from United States, and those different home markets shape how each company competes.

Meta Platforms, Inc.: Meta reported Q1 2026 revenue of $56.3 billion — up 33% year-over-year — with net income of $26.8 billion, up 61%. For a single quarter. Those figures imply an annualized revenue run rate exceeding $220 billion and a net income margin approaching 48%. The company had $201 billion in FY2025 revenue and $60.5 billion in net income. These are not the numbers of a company managing decline; they are the numbers of a company accelerating. Meta Platforms operates Facebook with 3.07 billion monthly active users, Instagram with more than 2 billion, WhatsApp with more than 2 billion, and Messenger, Threads, and the Quest virtual reality hardware line. The advertising system that monetizes this audience — auction-based, AI-optimized, targeting attention across six surfaces — generates 97.6% of the company's revenue. The remaining 2.4% comes from Reality Labs, the virtual reality and augmented reality division, which lost nearly $4 for every dollar it earned in FY2025. CEO Mark Zuckerberg controls the company through dual-class shares, giving him the authority to make decisions — including $125–145 billion in AI infrastructure investment in 2026 — without shareholder approval being a practical constraint. That capital program is one of the largest single-year corporate investment commitments in history and will determine whether Meta's AI capabilities remain competitive with OpenAI, Google, and the other systems competing for advertising-relevant AI capabilities. The company was founded as TheFacebook in February 2004 by Mark Zuckerberg and four Harvard classmates: Eduardo Saverin, Andrew McCollum, Dustin Moskovitz, and Chris Hughes. The Instagram acquisition in 2012 for $1 billion and the WhatsApp acquisition in 2014 for $22 billion are now recognized as two of the most consequential acquisitions in technology history, both completed well below what they would cost to recreate today.

OpenAI: That idealism would bend under the weight of economic reality. Training frontier AI models requires computational resources measured in the hundreds of millions of dollars per run. Its flagship product, ChatGPT, commands more than 300 million weekly active users as of early 2025. The free tier of ChatGPT, which offers access to GPT-4o mini and limited usage of GPT-4o, serves as the top of a carefully engineered conversion funnel. ChatGPT Plus, priced at $20 per month, unlocks priority access to the most capable models, image generation via DALL-E 3, web browsing, the ability to create and use custom GPTs, and — as of 2024 — access to memory features and voice capabilities. As of mid-2024, GPT-4o input tokens were priced at $5 per million and output tokens at $15 per million, while the more economical GPT-4o mini cost $0.15 per million input tokens and $0.60 per million output tokens. By early 2025, OpenAI claimed more than 92% of Fortune 500 companies were using its products in some form, though the depth of those engagements varied enormously from enterprise contracts to departmental API usage. OpenAI's Operator capability — announced in late 2024 — allows GPT-4o to take actions in web browsers autonomously, completing tasks like booking travel, filling forms, and managing software interfaces without human intervention. This positions OpenAI to capture transaction-layer economics rather than purely information-layer value. Gemini Ultra 1.0 reportedly outperformed GPT-4 on the MMLU benchmark across 57 academic subjects. However, Anthropic lacks OpenAI's consumer brand, its ChatGPT subscriber base, and the breadth of product surface area that allows OpenAI to capture multiple revenue streams simultaneously. Llama 3.1 405B, released in July 2024, was competitive with GPT-4 on several tasks and could be downloaded and run by any organization with sufficient GPU resources — at zero licensing cost. For OpenAI, the Llama series represents a price floor compression on API revenue; as open-weight models improve, price-sensitive API customers may migrate to self-hosted alternatives. While Stargate provides a path to the compute sovereignty OpenAI needs, it also represents a staggering capital commitment in a sector where the return timeline remains uncertain. Every conversation — corrected, upvoted, flagged, or refined — becomes training signal for subsequent model generations. The consumer flywheel is the first track. The nonprofit conversion faces scrutiny from California Attorney General Rob Bonta and Delaware courts examining whether existing investors are being treated equitably, a process that could take one to two years to resolve. The most strategically defining near-term product direction is AI agents: software that takes autonomous multi-step actions rather than generating single responses. If AGI were to emerge within a corporate context optimized for shareholder returns, who would ensure it was developed safely? The answer they arrived at was a nonprofit research laboratory with an open publication policy. The nonprofit structure would, in theory, ensure that decisions were made in the service of the mission rather than quarterly earnings. Sam Altman and Elon Musk served as co-chairs of the board. The early research agenda was ambitious and deliberately broad. OpenAI's founding team pursued work on reinforcement learning, robotics, natural language processing, and game-playing agents simultaneously, reflecting a conviction that AGI would likely emerge from the convergence of multiple models rather than any single architecture. By 2018, OpenAI Five, an enhanced version of the system, defeated professional human Dota 2 teams in exhibition matches watched by millions online. The research team also published the first version of the Generative Pre-trained Transformer — GPT-1 — in 2018, a language model trained on the BooksCorpus dataset of approximately 7,000 unpublished books. GPT-1 was not itself a commercial product; it was a research paper demonstrating that unsupervised pre-training on large text corpora could produce language representations transferable to downstream tasks. But it planted the seed for every commercial product that would follow. When that proposal was declined, and as Tesla's own AI efforts around autonomous driving created potential conflicts of interest, Musk resigned from the OpenAI board in February 2018. He would later claim in legal filings that he departed because he disagreed with the decision to pursue the capped-profit restructuring, and that he had been promised a different governance outcome. OpenAI disputes this characterization. The acrimony between Musk and OpenAI — particularly Altman — would become one of the defining interpersonal dramas of the AI industry. The decision was controversial internally and externally, with critics arguing it fundamentally compromised the organization's founding mission. The tension between these two positions has never fully resolved and remains the central fault line in OpenAI's institutional identity.

Meta Platforms, Inc. and OpenAI pursue distinct approaches to generating revenue, and understanding how each company operates is the foundation of any fair comparison between Meta Platforms, Inc. and OpenAI.

Meta Platforms, Inc. business model: Not subscriptions. Not commerce fees. Advertising sold through real-time auctions where millions of businesses bid against each other for attention slots in your feed, your Stories, your Reels, your inbox. The division loses nearly four dollars for every dollar it earns. Revenue model: Meta earns 97.6% of revenue from advertising sold across its Family of Apps — Facebook, Instagram, WhatsApp, Messenger, and Threads. ByteDance proved that algorithmic recommendation based purely on watch behavior could be more engaging than social-graph-based feeds. The competitive irony: TikTok invented the format, but Meta monetizes it better because it has the advertiser relationships, measurement infrastructure, and multi-surface distribution that ByteDance is still building. The multi-app strategy means behavioral shifts (from Feed to Stories to Reels to messaging) stay inside Meta's ecosystem rather than leaking to competitors. Short-form video now generates meaningful revenue as Meta has closed the gap between Reels ad loads and the more mature Feed and Stories surfaces. The format keeps growing in engagement, particularly on Instagram, and every percentage point of monetization parity with Feed represents billions in incremental revenue. That single rule — exclusivity by institutional trust — solved the identity problem that killed Friendster and made MySpace feel like a costume party. Chris Hughes shaped how the product communicated with students, making it feel like a campus utility rather than a tech startup's experiment.

OpenAI business model: The first and largest layer is consumer subscription revenue, centered almost entirely on ChatGPT. The consumer product's success is not merely a revenue story; it functions as the primary distribution channel for demonstrating model capability to potential enterprise buyers and developers, creating a virtuous cycle where consumer adoption subsidizes the feedback loops that improve model quality. Developers pay per token — units of text roughly equivalent to three-quarters of a word — with pricing tiered by model capability. Pricing is negotiated rather than published, but industry reporting suggests contracts range from $60 to $100 per user per month for larger deployments. The enterprise business is strategically critical because it generates predictable, recurring revenue from organizations with lower churn risk than individual consumers and because enterprise feedback loops accelerate fine-tuning and alignment work on models used in high-stakes professional contexts. Additionally, partnerships with companies like Morgan Stanley, which uses OpenAI models for wealth management research synthesis, and with healthcare organizations deploying GPT for clinical documentation, point toward a vertical-specialization revenue model where OpenAI captures premium pricing for domain-tuned AI applications. Leadership decisions about model release timing, pricing adjustments, and partnership structures are made against a background of competitive intelligence that changes weekly. Rather than competing on API pricing or enterprise features, Meta has pursued an open-weight model strategy with its Llama series that challenges the entire premise of proprietary AI as a defensible business. Meta's strategic logic is straightforward: the company spends billions annually on AI research as a cost center for improving its ad targeting and content recommendation systems, and releasing models as open-source creates an ecosystem that undermines competitors who monetize AI access as a product. Microsoft's Copilot products are built on OpenAI models today, but the company has been reportedly developing its own internal AI models — code-named MAI — that would reduce dependence on OpenAI in scenarios where the relationship deteriorates or pricing becomes unfavorable. In the United States, Federal Trade Commission scrutiny of the Microsoft-OpenAI relationship and the broader question of market concentration in foundation model APIs represents a long-term overhang. Competitive pressure from both sides — from well-capitalized incumbents like Google DeepMind and from fast-moving open-source alternatives like Meta's Llama family — poses an existential challenge to OpenAI's pricing power. The conversion funnel from free to Plus to Team to Enterprise is deliberately engineered: each pricing tier offers capability unlocks that make the next tier compelling to users who have already been habituated to AI assistance. By offering competitive pricing, extensive documentation, fine-tuning capabilities, and the custom GPTs marketplace, OpenAI aims to make its models the default infrastructure layer for AI application development — a position analogous to AWS for cloud computing. Finally, the autonomous agent track positions OpenAI for the next phase of AI monetization, where the company captures value not just for information generation but for task completion — a shift from a per-token pricing model to outcome-based or subscription-based pricing tied to measurable business results.

The durability of a company's moat often decides long-term winners. Here is how the competitive advantages of Meta Platforms, Inc. stack up against those of OpenAI.

Meta Platforms, Inc. competitive advantage: The 2026 capex guidance of $125-145 billion is almost entirely for AI infrastructure — NVIDIA H100 and H200 GPUs, custom silicon, and hyperscale data centers that will power recommendation algorithms, generative AI products, and the Llama model family. Meta wins on creative reach and audience scale. The AI infrastructure bet is staggering in scale. Network effects mean each new user makes the platform more valuable for existing users and advertisers. Is the advantage weakening? The most immediate payoff is Advantage+, Meta's AI-powered advertising suite. Everything depends on one variable: whether AI-generated revenue scales faster than AI infrastructure costs. Advantage+ is automating campaign creation and targeting so effectively that advertisers are spending more while doing less work. Llama models are becoming the default open-source foundation for enterprise AI development, which builds ecosystem lock-in without requiring Meta to charge licensing fees.

OpenAI competitive advantage: OpenAI's revenue architecture has evolved from a pure research-grant model into one of the most diversified monetization strategies in enterprise software, all built around a single core asset: access to frontier-scale artificial intelligence models. OpenAI's durable competitive advantages are fewer but deeper than those of most technology companies, and they derive from a combination of first-mover distribution scale, a uniquely advantaged compute infrastructure arrangement, and the compounding effects of the world's largest AI feedback dataset. The distribution moat is the most underappreciated advantage. ChatGPT's 300 million weekly active users as of early 2025 represent a data-generation engine of extraordinary scale. Anthropic, Mistral, and Cohere serve sophisticated enterprise users but lack the consumer scale that generates the breadth of conversational data needed to generalize across domains. By maintaining a generous free tier for ChatGPT, OpenAI accepts near-term revenue opportunity costs to maximize user scale, which in turn generates the preference data, usage patterns, and viral distribution that sustain model quality advantages. The developer ecosystem track recognizes that OpenAI's most durable moat is not its consumer brand but the millions of applications built on top of its API. Who would be accountable for its effects on labor markets, information ecosystems, national security, and individual autonomy? By publishing their research findings rather than hoarding them as trade secrets, they reasoned, they could accelerate the global scientific community's ability to understand and align advanced AI systems, reducing the advantage any single corporate actor could accumulate through secrecy.

Future prospects matter as much as current results. The growth strategies below explain how Meta Platforms, Inc. and OpenAI each plan to expand from here.

Meta Platforms, Inc. growth strategy: Under founder-CEO Mark Zuckerberg, Meta is investing $125-145B in AI infrastructure in 2026 alone — building massive GPU clusters to power recommendation algorithms, generative AI products (Meta AI assistant), and the Llama open-source model family. While they scroll, message, watch Reels, or browse Marketplace, Meta's AI systems build a behavioral profile so detailed that advertisers will pay premium prices to show those people specific ads at specific moments. The geographic revenue split reveals where the growth runway sits. The company is investing $125-145B in AI infrastructure in 2026. Strategic direction: AI-powered advertising automation (Advantage+), Reels monetization, WhatsApp business messaging, Meta AI assistant, Llama open-source models, Threads growth, and long-term Reality Labs investment in AR/VR computing platforms. In practice, neither is displacing the other — they're co-expanding the digital advertising market at the expense of television, print, and outdoor. Meta's response — Reels — now accounts for a growing share of time spent on Instagram and Facebook. Meta's counter-strategy is AI-powered conversion optimization and commerce tools like click-to-WhatsApp ads that create direct business conversations. Meta's ratio is almost double, and it's selling ads, not investment banking services. Most companies choose between growth and profitability. Investors looked at that number — larger than the annual revenue of all but about 30 companies on Earth — and asked: what exactly are the returns? The AI infrastructure means targeting and recommendation improve continuously, which improves engagement, which improves ad performance, which attracts more ad spend, which funds more AI investment. Meta's growth story in 2026 comes down to one word: AI. Not as a buzzword — as the literal engine driving every major initiative the company is pursuing. The honest assessment: Meta has two growth engines that matter right now (AI-powered ads and Reels) and two that could matter enormously in three to five years (WhatsApp commerce and AI assistants). If it does — and Q1 2026's 33% revenue growth on the back of Advantage+ suggests it might — then $125-145 billion in annual capex becomes the most profitable investment cycle since AWS. If it doesn't, Meta becomes a company spending like a sovereign wealth fund while growing like a utility. Viacom, Friendster's backers, various media executives: they all saw a college social network growing at a rate that made no commercial sense to leave independent. By spring 2004, TheFacebook had expanded to Columbia, Stanford, and Yale. Each campus launch followed the same playbook —.edu email gates, word-of-mouth virality, and the social pressure of being the last person in your dorm who hadn't signed up. Parker became Facebook's first president, introduced Zuckerberg to Peter Thiel, and helped secure a $500,000 angel investment that gave the startup room to breathe. The exclusivity that built trust was also a growth ceiling.

OpenAI growth strategy: The relationship would prove to be among the most consequential corporate partnerships in technology history. But the real story of OpenAI is less about personalities than about what happens when a small group of researchers actually builds something close to what they set out to build, and the world is not entirely sure it was ready for it. This usage-based pricing model scales elegantly with customer growth: as a developer's user base expands, their API consumption and therefore their OpenAI bill grow proportionally, creating a natural land-and-expand dynamic. The API business has high gross margins relative to infrastructure costs once models are trained, because the marginal cost of serving an additional API call decreases as batch sizes grow and inference optimization matures. The third layer, and the one commanding the most aggressive internal investment, is enterprise sales. The fourth layer, still emerging but strategically significant, encompasses Operator partnerships and vertical AI solutions. The ongoing and rapidly growing cost is inference: serving model outputs to hundreds of millions of users and API calls daily requires enormous and continuously expanding GPU clusters. At its operational core, OpenAI is an AI model development and deployment company whose product roadmap is determined by research breakthroughs rather than customer surveys. The organization is structured around research teams working on language models, multimodal systems, robotics (through a nascent hardware initiative), safety and alignment, and policy — with a product and go-to-market organization that translates research outputs into commercial applications. The pace of product releases has accelerated dramatically since ChatGPT's 2022 launch: in 2024 alone, the company released GPT-4o, GPT-4o mini, the Sora video generation model, real-time voice capabilities, the custom GPT store, and significant upgrades to DALL-E image generation. This dynamic creates an inherent tension in the partnership that neither side has publicly acknowledged but that shapes every major strategic decision. OpenAI's financial story in 2024 and 2025 is one of extraordinary revenue growth accompanied by equally extraordinary losses — a combination that defines the current phase of frontier AI development and raises genuinely difficult questions about when and whether the economics become sustainably profitable. The revenue growth trajectory implies a compound annual growth rate that has few parallels in enterprise software history. Compute costs have not fallen fast enough to offset the company's growth ambitions, and each successive generation of models requires exponentially more compute to train. Regulatory risk is expanding with the company's influence. OpenAI's growth strategy through 2027 rests on four parallel tracks that address different segments of the AI adoption curve simultaneously, each reinforcing the others through shared infrastructure, brand, and model improvement cycles. Expanding ChatGPT into mobile-first markets — the company's app is now available in over 160 countries and has been downloaded more than 500 million times — extends the consumer funnel into demographics where desktop PC penetration is lower but smartphone adoption is near-universal. The enterprise expansion track focuses on winning the largest and most regulated industries: financial services, healthcare, legal, and government. OpenAI's partnership with Morgan Stanley for financial advisor AI assistance, its collaborations with academic medical centers, and its early-stage discussions with government agencies through a nascent public sector division all point toward a deliberate verticalization strategy. This structure would unlock conventional equity compensation for employees, simplify the investor relationship, and create a cleaner path toward an IPO — which multiple sources have suggested could occur as early as 2026 depending on market conditions and the completion of regulatory reviews. OpenAI's Operator product and its broader agent framework suggest a future in which the company moves from selling access to intelligence to selling access to automated action — a shift that could expand the addressable market by an order of magnitude while also introducing new liability and regulatory considerations. The first notable public breakthrough came in 2017, when an OpenAI team developed Dota 2 playing agents that could defeat amateur human players in the complex strategy game — an achievement that demonstrated the potential of reinforcement learning in high-dimensional action spaces.

A closer look at the financial trajectory of Meta Platforms, Inc. and OpenAI rounds out the comparison.

Meta Platforms, Inc.: Revenue grew from $116.6 billion in FY2022 to $134.9 billion in FY2023, $201B in FY2025, and $201 billion in FY2025 — a four-year compound growth rate that few companies at this scale have sustained. Net income of $60.5 billion in FY2025 represents a 30% net margin on a $201 billion revenue base, an extraordinary result for an advertising business. The 2022 revenue dip was driven by two simultaneous pressures: Apple's App Tracking Transparency update, which degraded the targeting signal Meta's advertisers depended on, and macroeconomic softness in digital advertising spend. The company recovered through AI-powered targeting models that reconstructed purchase intent signals from less granular data, and through AI-driven feed and Reels optimization that increased engagement duration and therefore inventory yield. The $125–145 billion AI infrastructure investment planned for 2026 is the most aggressive capital commitment in Meta's history and one of the largest annual capex programs of any company globally. This investment funds data centers, custom AI chips, and the infrastructure to train and serve the models that power content ranking, ad targeting, and generative AI products. The commercial return on this investment will be measured in advertising CPMs and engagement minutes, not in direct AI product revenue. Reality Labs generated approximately $900 million in FY2025 revenue while losing close to $4 billion. The cumulative losses from Reality Labs since 2019 exceed $40 billion. Zuckerberg has described this as a generational bet. The financial discipline that allows a $40 billion loss in one division while generating $60 billion in net income overall is only possible because the Family of Apps advertising business is structurally exceptional.

OpenAI: OpenAI was incorporated in December 2015 as a nonprofit research laboratory in San Francisco, funded by an initial $1 billion pledge from a group of investors and technologists that included Elon Musk, Peter Thiel, Reid Hoffman, and a young Sam Altman. By 2019, OpenAI created a subsidiary with a 'capped-profit' structure — limiting investor returns to one hundred times their investment — and accepted a $1 billion investment from Microsoft. By 2023, Microsoft had deepened that commitment to approximately $13 billion across multiple tranches, embedding OpenAI's technology into virtually every major Microsoft product from Word and Excel to GitHub and Azure cloud services. By fiscal year 2024, OpenAI was generating an annualized revenue run rate exceeding $3.7 billion, a figure that climbed with stunning velocity toward an estimated $5 billion in full-year 2024 revenue, with projections pointing toward $11.6 billion in 2025. Those numbers arrived alongside staggering costs: the company reportedly spent more than $7 billion in 2024 alone, with compute bills from running inference on hundreds of millions of ChatGPT queries contributing to operating losses that were expected to narrow only as model efficiency improved. Despite the losses, investors in late 2024 valued OpenAI at $157 billion in a funding round that raised $6.6 billion — and by early 2025, secondary market transactions and strategic discussions suggested a valuation exceeding $300 billion, placing it among the most valuable private companies in American history. The company generated an estimated $5 billion in revenue in 2024, driven by ChatGPT subscriptions, API access for developers, and enterprise contracts, with 2025 revenue projected at $11.6 billion. Microsoft has invested approximately $13 billion in the company and distributes OpenAI models through Azure OpenAI Service. With a reported valuation of $300 billion and competition intensifying from Google DeepMind, Anthropic, Meta AI, and xAI, OpenAI sits at the center of the most consequential technology race of the twenty-first century. By late 2024, OpenAI had approximately 15 million paying ChatGPT subscribers, generating estimated annualized revenue of roughly $2 billion from this segment alone. Microsoft's $13 billion investment did not flow to OpenAI as cash in the conventional sense; a significant portion was structured as Azure cloud credits, meaning OpenAI receives the compute it needs to train and serve models at scale without cash outlays, while Microsoft receives a percentage of OpenAI's revenue and exclusive rights to commercialize OpenAI technology outside of OpenAI's own products. Model training costs for a single frontier model run — GPT-4 reportedly cost over $100 million to train — are capital-intensive one-time expenditures. In 2024, OpenAI's total operating costs were estimated at more than $7 billion, driven primarily by compute, personnel — with AI researchers commanding packages in the millions of dollars — and safety and alignment research teams. The company operates at a substantial net loss by conventional accounting, with losses reportedly exceeding $5 billion in 2024, though the trajectory of margin improvement is steep as inference efficiency gains from techniques like speculative decoding, quantization, and custom silicon accumulate. Looking at the unit economics differently: OpenAI's 2024 revenue of approximately $5 billion against roughly 3,500 employees implies revenue per employee of approximately $1.4 million — already among the highest in the software industry. As the company scales revenue toward its projected $11.6 billion in 2025 without proportional headcount growth, the leverage in the model becomes visible. OpenAI is a Artificial Intelligence / Technology company with $5B in 2024 revenue and 4K employees worldwide. Anthropic has raised more than $7.3 billion, including a $4 billion commitment from Amazon and a $2 billion commitment from Google, and its Claude 3.5 Sonnet model received widespread recognition in 2024 for outperforming GPT-4o on several coding and reasoning benchmarks. Grok 2, released in mid-2024, demonstrated genuine capability improvements, and xAI's December 2024 funding round at a $50 billion valuation signaled that investors viewed the venture as a credible tier-one AI lab. The company generated an estimated $3.7 billion in annualized revenue by the end of 2024's third quarter, with full-year 2024 revenue reaching approximately $5 billion according to multiple reporting sources including The Wall Street Journal and The New York Times. That figure represented roughly threefold growth from 2023 revenues estimated at $1.6 billion, themselves a dramatic increase from the sub-$30 million the company earned in 2022 before ChatGPT launched. Against that revenue, operating costs in 2024 were estimated at more than $7 billion, producing an operating loss of approximately $5 billion. The largest cost components were compute infrastructure, AI researcher compensation — top researchers reportedly earn total packages of $3 million to $10 million annually — and safety and policy staff. The company's runway was extended substantially by its October 2024 funding round, which raised $6.6 billion at a $157 billion post-money valuation from investors including Thrive Capital, SoftBank, Fidelity, and others. Looking forward, OpenAI's own internal projections, reported by The Financial Times and Bloomberg, call for 2025 revenues of $11.6 billion and project a path to profitability around 2029, contingent on model efficiency improvements that reduce per-query compute costs and continued growth in the enterprise subscriber base. The Stargate infrastructure joint venture, if executed at its announced $500 billion scale over four years, would fundamentally alter the company's compute cost structure by internalizing infrastructure that is currently expensed as operating cost. OpenAI lost an estimated $5 billion in 2024, a figure that reflects the brutal economics of training and serving frontier AI at scale. The company has publicly discussed spending $500 billion on AI infrastructure through the Stargate project, a joint venture with SoftBank and Oracle announced by President Donald Trump in January 2025. The Stargate project, announced in January 2025 with President Trump present at the announcement, envisions $500 billion in AI infrastructure investment over four years through a joint venture involving OpenAI, SoftBank, and Oracle. The primary concern at the time was Google's acquisition of DeepMind in 2014 for approximately $625 million and its subsequent acquisition of multiple other AI research groups. The same year, facing the computational reality that training ever-larger models required capital that a nonprofit simply could not raise, the board approved the creation of the OpenAI LP subsidiary — the capped-profit entity — and accepted Microsoft's first $1 billion investment.