OpenAI's financials & how to track them

Microsoft is OpenAI's largest strategic backer and the company's primary cloud partner through Azure. It holds roughly a 27% stake in OpenAI, hosts training and inference workloads, and embeds OpenAI models across its product stack, from Copilot in Office and Windows to GitHub Copilot and Azure OpenAI Service. When OpenAI raises capital or commits to new compute, Microsoft's balance sheet and Azure revenue are often on the other side of the trade. For public-market investors who want the closest large-cap proxy to OpenAI's trajectory without waiting for an IPO, MSFT is the default starting point.

AMD is the main alternative AI accelerator supplier as hyperscalers and frontier labs diversify away from a single GPU vendor. Its MI300 and follow-on Instinct chips target the same training and inference workloads as Nvidia, often at a lower price point per flop. Microsoft, Meta, and others have publicly committed to AMD clusters alongside Nvidia racks. The stock offers a second way to own AI compute silicon when you believe demand is large enough for more than one winner.

Micron is one of the world's largest memory makers and a growing supplier of high-bandwidth memory (HBM), the specialized DRAM stacked alongside AI GPUs. Training clusters are increasingly limited by memory bandwidth, not just raw FLOPs, so every new Nvidia generation pulls more HBM per server. Micron has been ramping HBM capacity to meet hyperscaler orders tied to AI build-outs. The stock gives you the picks-and-shovels layer one step below the GPU vendors themselves.

TSMC fabricates the advanced processors behind the AI boom, including Nvidia GPUs, AMD accelerators, and custom chips from hyperscalers. It operates at the leading edge of semiconductor manufacturing and is the foundry of choice when designers need the densest transistors available. AI demand has kept TSMC's most advanced nodes heavily booked, with long wait times for allocation. Owning TSMC is a bet on AI chip volume regardless of which accelerator architecture wins.

CoreWeave is a GPU-focused cloud built for AI training and inference, with direct contracts to frontier labs including OpenAI. It rents Nvidia clusters by the hour instead of selling chips, so revenue scales with how much compute the industry burns. The company went public amid intense demand for GPU capacity that hyperscalers alone could not satisfy. Among US-listed names, CoreWeave is one of the purest ways to own AI compute rental rather than the hardware or software layers above and below it.

Alphabet is both a backer of OpenAI's rival Anthropic and a direct competitor through its Gemini models and Google Cloud AI services. Google has committed billions to Anthropic while racing to ship its own frontier systems and TPUs. That dual role makes GOOGL a useful way to track the competitive side of the AI race: whoever wins between OpenAI and Anthropic, Google is still positioned across chips, cloud, and consumer distribution. It is not an OpenAI proxy, but it completes the picture for investors following the full landscape.

SpaceX completed a record-setting IPO in 2026 and now sits at the intersection of launch services, Starlink, and frontier AI. In early 2026 it acquired xAI, bringing the Grok models in-house as SpaceX's AI division. That makes SpaceX one of the few publicly tradable tickers with a direct line into a frontier lab, not just the hardware or cloud layers underneath. Investors are effectively buying a bundle of space infrastructure and AI model capability in a single name. For OpenAI watchers, SpaceX is a comparable case study in how the market might treat a blockbuster AI-related listing when shares finally trade.