What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
NVIDIA is currently the clearest large-company winner from AI infrastructure spending. Microsoft is the clearest major software-and-cloud company to disclose a substantial AI-specific revenue run rate. Alphabet, Amazon, Meta, Oracle, Broadcom, TSMC and AMD also benefit, but their AI economics are often bundled with broader businesses or reported only as indirect exposure.
There is no authoritative league table of AI revenue. Companies use different fiscal years and rarely separate AI sales, costs and profits from conventional products. The ranking below therefore distinguishes direct AI sales, AI-exposed infrastructure and AI-enabled businesses instead of pretending unlike figures are directly comparable.
What “making money from AI” actually means
AI-related business falls into four different categories:
- Direct AI revenue: AI accelerators, model APIs, AI subscriptions, software licences and explicitly sold AI cloud services.
- AI-exposed revenue: cloud capacity, networking, memory, chip manufacturing and data-centre equipment used for AI workloads alongside conventional workloads.
- AI-enabled revenue: advertising, search, recommendations, commerce or productivity products whose economics improve because of machine learning or generative AI.
- Investment and valuation: capital expenditure, backlog, market value and funding. These may signal expectations, but they are not revenue or profit.
A dollar can also appear at several layers. An enterprise buys an AI application; its developer buys cloud capacity; the cloud provider buys GPUs, networking and data-centre equipment; and the chip designer pays a foundry. Adding every layer would count the same end-market spending repeatedly.
#1 Best Overall
The leading companies by monetisation clarity
| Company | Main AI money engine | Best disclosed figure | What the figure is | Profitability or risk signal |
|---|---|---|---|---|
| NVIDIA | Data-centre GPUs, systems, networking and AI software | $215.9 billion fiscal 2026 revenue | Total company revenue, not pure AI revenue | Largest and most direct public-company exposure; includes non-AI businesses and is cyclical |
| Microsoft | Azure AI, Copilot, GitHub and enterprise applications | More than $37 billion AI annual revenue run rate | Management-defined annualised run rate, not an audited segment | Microsoft Cloud gross margin was 66% in fiscal Q3 2026; infrastructure investment weighs on margins |
| Alphabet | AI-enabled search and advertising, Google Cloud and Gemini | AI revenue not separately disclosed | Search, advertising and cloud are broader segments | Enormous installed businesses monetise AI indirectly |
| Amazon | AWS capacity, Bedrock, custom chips, advertising and retail optimisation | AI revenue not separately disclosed | AWS and commerce figures include conventional activity | Scale and recurring cloud consumption; AI cannot be isolated |
| Meta | AI-improved advertising, recommendations and consumer tools | AI revenue not separately disclosed | AI is mainly embedded in advertising and engagement | Large infrastructure spending is an investment, not revenue |
| Oracle | Oracle Cloud Infrastructure and enterprise AI workloads | Negative $23.7 billion fiscal 2026 free cash flow | Company-wide cash flow, not AI revenue | Cloud growth is paired with an aggressive, capital-intensive buildout |
| Broadcom | Custom accelerators, Ethernet switching and connectivity | AI-specific total not stated | Company reporting combines AI and non-AI semiconductor activity | Benefits as hyperscalers design custom chips, but depends on a few large customers |
| TSMC | Manufacturing advanced AI chips and packaging | AI-specific total not stated | Foundry revenue across customers and applications | Manufacturing bottleneck with very high capital requirements |
| AMD | Instinct accelerators, EPYC CPUs and data-centre platforms | $34.6 billion 2025 revenue; $16.6 billion data-centre revenue | Data-centre revenue includes CPUs and other products | Visible accelerator competitor, but the proxy is not pure AI sales |
1. NVIDIA: the clearest direct AI infrastructure winner
NVIDIA reported $215.9 billion in fiscal 2026 revenue, with growth driven by accelerated computing and AI. Its data-centre business supplies GPUs, complete systems, networking and interconnects to hyperscalers, cloud providers, laboratories, enterprises and governments. The company also sells AI software, enterprise platforms, automotive products and physical-AI systems.
This makes NVIDIA the strongest answer when “making money from AI” means selling the infrastructure required to train and run models. Its fiscal fourth-quarter revenue was $62.3 billion, another indicator of the scale of current demand. The figures are reported in NVIDIA’s filings and results releases (fiscal 2026 filing; quarterly results).
That is not the same as saying all $215.9 billion was AI revenue. Accelerated computing includes workloads beyond generative AI, while gaming, automotive, networking and enterprise sales are not wholly AI businesses. The defensible description is that NVIDIA is the largest and most directly exposed public-company beneficiary of AI infrastructure spending.
2. Microsoft: the broadest disclosed software-and-cloud monetiser
Microsoft said in fiscal Q3 2026 that its AI business had exceeded a $37 billion annual revenue run rate. That measure is annualised and company-defined, not a separately audited GAAP segment, so it should not be compared mechanically with NVIDIA’s fiscal-year revenue.
Microsoft monetises AI at several layers: Azure infrastructure and model hosting, Microsoft 365 Copilot, GitHub Copilot, Dynamics, security, data services and enterprise applications. Microsoft Cloud revenue was $54.5 billion in the quarter, and Azure and other cloud services grew 40% year over year. More than 20 million paid Microsoft 365 Copilot commercial seats were reported. These figures are available in Microsoft’s earnings release, metrics and earnings-call materials.
Azure growth includes non-AI workloads, and Microsoft Cloud includes products unrelated to AI. Profitability also matters: Microsoft Cloud’s fiscal Q3 2026 gross margin was 66%, with higher AI usage and infrastructure investment reducing the percentage (performance details).
Alphabet, Amazon and Meta: AI-enabled businesses
Alphabet
Alphabet earns from AI through search advertising, AI-assisted ad tools, Google Cloud infrastructure and model services, Gemini subscriptions and YouTube recommendations and advertising. AI Overviews and Gemini can change how users search, but Alphabet does not publish a consolidated AI-revenue line. Search revenue cannot be labelled entirely AI revenue, and Google Cloud revenue is not equivalent to AI revenue. Alphabet’s investor materials are available at Alphabet Investor Relations.
Amazon
Amazon monetises AI through AWS compute, storage, networking, Bedrock model access and custom Trainium and Inferentia chips. It also uses AI in advertising, product discovery, fulfilment and customer service. AWS remains a very large conventional-cloud business, and Amazon has not provided a comprehensive AI-revenue figure. Its financial reporting is available through Amazon Investor Relations.
Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Scan for outdated or missing drivers - takes under a minute3Clear out junk files and repair common Windows errorsRank #3
Meta
Meta’s financial payoff is primarily indirect: AI improves ad targeting, ranking, creative generation, measurement, recommendations, engagement and retention across Facebook and Instagram. Meta AI, business messaging and automated customer-service tools add direct products, while custom infrastructure is mostly an internal capability. Large AI capital expenditure is therefore not evidence of current AI revenue, and no clean consolidated AI-revenue figure is reported here.
Critical suppliers behind the AI buildout
Broadcom
Broadcom supplies custom application-specific chips, Ethernet switching, connectivity and data-centre components. Hyperscalers increasingly design their own accelerators, creating demand for Broadcom’s design and networking expertise. Its total revenue also includes substantial non-AI semiconductor and infrastructure-software activity; current company disclosures should be consulted at Broadcom Investor Relations.
TSMC
TSMC manufactures advanced logic chips and packaging for designers including NVIDIA, AMD and others. It benefits from demand across competing AI-chip architectures, but foundry revenue is not the same as revenue from selling an AI service to an end user. TSMC’s disclosures are at TSMC Investor Relations.
AMD
AMD reported $34.6 billion of 2025 revenue and $16.6 billion of data-centre revenue, up 32% year over year. Data-centre sales include EPYC CPUs and other products as well as Instinct accelerators, so the $16.6 billion figure is an AI-exposed proxy rather than pure AI revenue (AMD filing).
Quick wins for a faster PC:
Clear out junk files and repair common Windows errorsFree Scan →Scan for outdated or missing drivers - takes under a minuteDriver Scan →Revenue is not the same as AI profit
To judge whether AI is creating economic value, examine revenue, gross profit, operating profit, free cash flow and return on invested capital together.
- Chip designers can achieve high gross margins, but face product cycles, supply constraints and customer concentration.
- Cloud providers gain recurring consumption revenue, while depreciation, power, networking and data-centre construction absorb cash.
- Software companies may have strong subscription economics, but inference, research and model-access costs can reduce incremental margins.
- Advertising platforms can improve revenue without selling a separately priced AI product.
- Foundries benefit from wafer demand but must continually fund enormous capacity investments.
Oracle is a useful warning against equating growth with cash generation: it reported negative $23.7 billion of fiscal 2026 free cash flow while expanding cloud infrastructure (results; annual filing). A company can therefore have strong AI-related demand and weak current free cash flow.
Who is monetising now, and who is still investing?
Clearly monetising current demand
- NVIDIA through data-centre accelerators and systems.
- Microsoft through Azure, Copilot and enterprise AI products.
- AWS, Google Cloud and Oracle Cloud through AI-exposed consumption.
- Alphabet and Meta through AI-enabled advertising and recommendations.
- Broadcom, TSMC and AMD through chips, networking and manufacturing.
More dependent on future returns
- Meta’s large infrastructure programme, where much of the capability supports internal products.
- Oracle’s capacity expansion, which requires substantial upfront spending.
- Private model developers and startups whose revenue may be unaudited, annualised or still dependent on external capital.
How to compare AI companies without false precision
- Identify the monetisation layer: direct product, cloud infrastructure, supplier, or AI-enabled legacy business.
- Label the number: audited revenue, segment proxy, annualised run rate, management estimate or analyst estimate.
- Match periods: NVIDIA’s fiscal year ends in January, AMD’s 2025 year ended December 27, Microsoft’s fiscal year ends June 30 and Oracle’s ended May 31, 2026.
- Check cash economics: compare gross margin, operating income, free cash flow and capital expenditure.
- Adjust for concentration and durability: assess dependence on a few hyperscalers, temporary equipment shortages, model-price declines and commoditisation.
- Avoid double counting: a cloud provider’s AI revenue and a GPU supplier’s sales may come from the same customer spending.
Market capitalisation, stock performance, bookings, backlog and remaining performance obligations can reflect expectations, but none is a substitute for realised AI revenue or profit.
What about OpenAI, Anthropic and other private AI companies?
Private laboratories belong in a separate comparison. Their revenue figures may be unaudited or presented as annualised run rates, and rapidly growing sales can coexist with negative margins because inference and training are expensive. They also buy compute from the same cloud and chip companies that benefit from their expansion. Any private-company figure should be attributed to the company or a named reporting source rather than treated as directly comparable with public-company filings.
The bottom line
NVIDIA is the largest direct AI-infrastructure monetiser on the evidence available, while Microsoft has the strongest disclosed large-company AI run rate spanning cloud and software. Alphabet and Meta monetise AI mainly through advertising, search and engagement; Amazon, Google Cloud, Microsoft Azure and Oracle sell AI-exposed cloud capacity; Broadcom, TSMC and AMD supply the hardware and manufacturing base. The unresolved investment question is not whether AI generates revenue, but whether that revenue will ultimately exceed the industry’s enormous costs for compute, power, talent and data centres.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
