Which technology companies are furthest along with AI?

In the Enterprise AI diffusion report, Alphabet, Microsoft, and Salesforce are at Scaling, Accenture, Amazon, IBM, Intuit, Meta Platforms, NVIDIA, and Shopify are at Piloting, and Apple, Oracle, and Samsung Electronics are at Experimenting. A stage follows from the returns and rollouts a company has publicly reported in the last 18 months, not from its spending or plans.

How do tech companies use AI internally?

Technology companies use AI internally most in software engineering, customer support, sales, IT and HR. Microsoft credits AI with about $750 million a year of customer service savings (Apr 2026), Salesforce runs its Help Agent on help.salesforce.com and says AI coding tools doubled the features and code it shipped with flat engineering headcount (May 2026), Alphabet says about 75% of Google's new code is AI-generated (Apr 2026) and 83% of its sales team uses Gemini-assisted tools weekly (Jul 2026), and Meta Platforms engineers write code with Meta's own MetaCode and Muse Code while the REA agent runs ads ranking experiments (Mar 2026). Amazon credits its Rufus shopping assistant with nearly $12 billion in incremental annualized sales in 2025 (Feb 2026), says 83% of its engineers use its Kiro coding tool (May 2026), and gives staff internal assistants such as Quick, Aza and MeshClaw. IBM's AskIT agent serves about 270,000 employees, and NVIDIA's IT organization runs an internal AI factory for about 40,000 employees (Jun 2026). Accenture is rolling out Microsoft 365 Copilot to about 743,000 employees (Apr 2026), Shopify's River agent in Slack touched the work of more than 7,000 people in 30 days (May 2026), Intuit runs AI agents in its customer success operations beside human tax and bookkeeping experts, and Samsung Electronics staff use its own Gauss Portal assistant and code.i coding tool.

What return have technology companies reported from AI?

Technology companies report AI returns in a few functions and no firm-level figure. Amazon credits Rufus, its AI shopping assistant, with nearly $12 billion in incremental annualized sales in 2025 (Feb 2026). Microsoft credits AI with about $750 million a year of customer service savings (Apr 2026), an estimated $10 million a year from a supplier sourcing assistant (Jun 2025) and about 50% lower IT asset-management time and costs (Jan 2026). Salesforce credits Agentforce with about $100 million taken out of support costs (Oct 2025) and AI coding tools with doubling the features and code shipped at flat engineering headcount (May 2026). Alphabet credits Gemini tools for its ad sales teams with hundreds of millions of dollars of incremental revenue and over 10% higher seller productivity (Oct 2025), and estimates that AI raised engineering velocity about 10% (Jun 2025). Meta Platforms credits AI, mostly agentic coding, with a 30% rise in output per engineer since the start of 2025 (Jan 2026), and Intuit credits generative AI tools with 39% more code delivered per developer (Dec 2025). IBM credits AI agents with replacing the work of a couple hundred HR staff (May 2025). Accenture, NVIDIA, Shopify, Apple, Oracle and Samsung Electronics report no business return they credit to AI alone.

Do tech companies use their own AI products internally?

Technology companies that make AI mostly run it on themselves first. Microsoft deploys Microsoft 365 Copilot, Copilot Studio, Foundry, GitHub Copilot and Agent 365 under its Customer Zero program, Salesforce runs Agentforce and Slackbot on itself under its own Customer Zero program, IBM builds its internal AI on watsonx, Alphabet runs Google on its own Gemini models, TPUs and Antigravity agent platform, Amazon's engineers and staff use Kiro and Quick, AWS products Amazon also sells, Meta Platforms builds tools such as MetaCode, Muse Code and Metamate on its own Muse and Llama models, and NVIDIA's internal AI factory runs on NVIDIA's own products. Intuit built its own GenOS platform and financial language models, and Samsung Electronics runs its own Samsung Gauss models on its internal network. Oracle bought ChatGPT Enterprise and Codex from OpenAI for its employees (Apr and May 2026), Accenture is rolling out Microsoft 365 Copilot (Apr 2026), and Apple uses its own Foundation Models with Anthropic Claude and Google Gemini, as reported.

Which AI models do tech companies use internally?

Technology companies mostly combine their own models with models from OpenAI, Anthropic and Google. Alphabet uses Gemini by default and has offered Anthropic's Claude Opus 5 to every Google engineer inside Antigravity on a per-user quota since Sep 2026, while Meta Platforms runs its own Muse and Llama models and has steered engineers from Claude Code to its own MetaCode, with Claude Code users falling from about 60,000 to about 30,000 (The Information, Oct 2026). Amazon uses its own Nova models alongside Anthropic's Claude and OpenAI models, and opened Claude Code and Codex to all builders through Bedrock (May 2026). Intuit pairs its own financial language models with OpenAI, under a contract worth more than $100 million (Nov 2025), Anthropic and Amazon Nova, and Samsung Electronics pairs its own Samsung Gauss models with ChatGPT, Gemini Enterprise and Claude for Device eXperience staff (Jun 2026). Salesforce made Claude Code the primary coding agent for all its engineers (May 2026), Microsoft uses OpenAI models by default plus Anthropic and its own MAI models, IBM runs Claude, Mistral and Granite in its Bob coding agent, and Oracle gives staff OpenAI's ChatGPT and Codex (Apr and May 2026).

How technology companies deploy AI

Technology companies credit AI with returns in selling, support and code, and AI now sits on both sides of their cost line. Amazon credits its Rufus shopping assistant with nearly $12 billion of incremental annualized sales, Microsoft credits AI with about $750 million a year of customer service savings, and Salesforce doubled the features and code it ships with engineering headcount mostly flat. The bill rises with use: NVIDIA's CFO named AI tool use as a driver of expense growth, and Salesforce expects about $300 million of Anthropic token spend, mostly for coding.

Verified Nov 5, 2026See methodology

How technology companies have deployed AI

Findings

What the technology profiles show

Sales and customer support are where technology companies report dollar returns from AI: Amazon credits Rufus, its AI shopping assistant, with nearly $12 billion in incremental annualized sales in 2025 (Feb 2026), Microsoft credits AI with about $750 million a year of customer service savings (Apr 2026), Salesforce credits Agentforce with about $100 million taken out of support costs (Oct 2025), and Alphabet says Gemini tools for Google's ad sales teams raised seller productivity by over 10% and led to hundreds of millions of dollars in incremental revenue (Oct 2025). Microsoft also credits a supplier sourcing assistant with an estimated $10 million a year (Jun 2025) and the myDevice AI agent with about 50% lower IT asset-management time and costs (Jan 2026), and IBM credits AI agents with replacing the work of a couple hundred HR staff (May 2025). No technology company reports a firm-level figure it credits to AI, and none is at Embedded.
Software engineering is where the most technology companies credit AI with a return: Salesforce said in May 2026 that AI coding tools doubled the features and code shipped year over year while engineering headcount stayed mostly flat at about 15,000, Meta Platforms CFO Susan Li said in Jan 2026 that output per engineer rose 30% since the start of 2025, mostly from agentic coding, Intuit credits generative AI tools with 39% more code delivered per developer (Dec 2025), and Alphabet estimates that AI raised engineering velocity about 10% across the company (Jun 2025). Other technology companies report coding results as operating measures with no cost or headcount figure: Shopify says AI writes well over half its code (May 2026), Cursor published 3 times more committed code at NVIDIA (Feb 2026), IBM counts more than 80,000 employees on its Bob coding agent (Apr 2026) without a return credited to Bob, Amazon says 6 engineers using Kiro rebuilt Bedrock's inference engine in 76 days, work it says would normally take about 40 people a year (Apr 2026), and Samsung Electronics reports that design agents cut HBM4 design lead time by 50% (Mar 2026).
Technology companies that make AI run it on themselves before customers get it: Microsoft deploys Microsoft 365 Copilot, Copilot Studio, Foundry, GitHub Copilot and Agent 365 internally under its Customer Zero program, Salesforce has run Agentforce on help.salesforce.com since Oct 2024 and Slackbot since Jan 2026 under its own Customer Zero program, and IBM builds its internal AI on watsonx Orchestrate, watsonx.data and watsonx.governance (Jan 2026). Sundar Pichai wrote in Apr 2026 that Google always wants to be customer zero for its own technologies, and Alphabet runs its internal AI on the Gemini models, TPUs and Antigravity agent platform it sells. Amazon builds its internal AI tools on AWS, where Kiro for coding and Quick as the work assistant are AWS products Amazon also sells, and serves Claude Code and Codex to its builders through Bedrock (May 2026). NVIDIA, the semiconductor maker, runs an internal AI factory built on NVIDIA's own products that serves about 40,000 employees (Jun 2026), and Meta Platforms builds most of its internal tools, such as MetaCode and Muse Code for engineers and Metamate for staff, on its own Muse and Llama models. Oracle instead bought its employee AI tools from OpenAI, ChatGPT Enterprise and Codex (Apr and May 2026); Shopify built its own layer, an internal LLM proxy, Aquifer, River and Quick, while buying Cursor and Claude Code; Intuit built GenOS, its own generative AI platform (Jun 2023), with its own financial language models beside commercial ones; and Samsung Electronics built Samsung Gauss after engineers leaked source code to ChatGPT in 2023, then added ChatGPT, Gemini Enterprise and Claude for its Device eXperience staff (Jun 2026).
Technology companies are moving their engineers in opposite directions on Anthropic's Claude: Salesforce made Claude Code the primary coding agent for all its engineers and removed token limits (May 2026); Intuit has deployed Claude Code across its engineering organization since Feb 2026; Amazon, after a Nov 2025 memo told engineers to standardize on its own Kiro, opened Claude Code and OpenAI's Codex to all builders in May 2026 once about 1,500 employees backed formal Claude Code adoption; Alphabet opened Claude Opus 5 to every Google engineer inside Antigravity on a per-user quota, while Claude Code and OpenAI's Codex stay barred for most staff (Sep 2026); and Meta Platforms steered engineers to its own MetaCode and Muse Code, with Claude Code users falling from about 60,000 earlier in 2026 to about 30,000 (Oct 2026). The Alphabet and Meta Platforms figures come from Business Insider and The Information, not from the companies.
Technology companies tie AI to workforce and cost changes without saying how much AI contributed: Oracle's Jun 2026 Form 10-K says AI adoption resulted in workforce reductions as staff fell from about 162,000 to 141,000 in fiscal 2026, without an AI share; IBM's more than $5.5 billion of productivity (Oct 2026) blends AI with automation, process simplification and spend cuts; Accenture's revenue per person rose about 5% in fiscal 2026 partly from delivery efficiencies, with AI's share not separated (Oct 2026); Intuit's expected $135 million or more of fiscal 2026 efficiencies is credited to AI and automation together; and Shopify credits its falling operating expenses as a share of revenue mainly to headcount discipline (Feb 2026). Andy Jassy told Amazon staff in Jun 2025 that generative AI would shrink its corporate workforce, and Amazon cut about 14,000 corporate roles in Oct 2025 and 16,000 in Jan 2026, while Jassy said the first round was not really AI-driven. Microsoft said its Jul 2026 job cuts were not replaced by AI, Intuit's Sasan Goodarzi said its May 2026 cut of about 17% of staff had nothing to do with AI, Meta Platforms has not credited AI with the about 8,000 jobs it cut from May 2026, and Alphabet's headcount grew 7.7% in the year to Sep 2026.
Technology companies now name their own AI use as a cost: NVIDIA's Colette Kress named rising AI tool use as a driver of operating expense growth in fiscal 2027 (Aug 2026) without saying what the tools return, Meta Platforms named third-party AI token costs among the drivers of its expense growth in Q2 and Q3 2026 and told staff it would manage tokens with budgets from 2027 (Jun 2026), and Salesforce's Marc Benioff expects about $300 million of Anthropic token spend in 2026, mostly for coding (May 2026). Alphabet rations compute for internal AI: engineers using internal coding tools often hit capacity limits (Apr 2026), and Claude Opus 5 comes on per-user quotas (Sep 2026). Amazon shut down KiroRank, an employee-built token leaderboard, in May 2026, and its revised software engineer role guideline calls token usage and AI tool adoption deeply flawed if used as targets (Oct 2026).
Apple, Oracle and Samsung Electronics are at Experimenting because none of them has officially reported how many employees use AI: what is known of Apple's internal AI, such as the Enchanté assistant rolled out to office staff from Nov 2025, comes from press reports (Jan 2026); Oracle's user counts come from Business Insider's account of a town hall (Sep 2026) and OpenAI's count of 130,000 active ChatGPT users (Oct 2026); and Samsung Electronics opened ChatGPT, Gemini Enterprise and Claude to all Device eXperience staff (Jun 2026) without Samsung or its vendors saying how many use them. Meta Platforms, Amazon and Intuit have not published company-wide counts either and are at Piloting on their returns: Meta Platforms' Claude Code and MetaCode user counts come from The Information (Oct 2026), Amazon says 83% of its engineers use Kiro (May 2026), and Intuit says a majority of its workforce uses AI tools regularly (Dec 2025). Alphabet reports adoption by group: about 75% of new code is AI-generated (Apr 2026) and 83% of the sales team uses Gemini-assisted tools weekly (Jul 2026). Other technology companies report their own rollouts with counts: Salesforce says 99% of its workforce used internal agents (Sep 2025), Accenture reports 89% monthly active use in a tranche of about 200,000 Microsoft 365 Copilot licenses (Apr 2026), IBM more than 80,000 employees on Bob (Apr 2026), Microsoft 58,000 active Copilot Cowork users (Aug 2026), NVIDIA more than 10,000 employees across functions on Codex (Apr 2026), and Shopify 6,273 employees in River threads in Sep 2026.

Stages

Where technology companies stand

Scaling
Alphabet, Microsoft, Salesforce
Scaling means the company reports returns it credits to AI in several functions, or a firm-level AI figure, and runs a company-wide AI platform with a named owner.
Piloting
Accenture, Amazon, IBM, Intuit, Meta Platforms, NVIDIA, Shopify
Piloting means the company reports a business return it credits to AI in at least one use case, or a company-wide AI rollout with a user count and a named owner.
Experimenting
Apple, Oracle, Samsung Electronics
Experimenting means the company has reported neither a business return it credits to AI nor a company-wide AI rollout with a user count. Pilots, partnerships, plans and spending alone keep a company at Experimenting.

Returns

The returns technology companies credit to AI

Only realized returns the company reported and credited to AI itself, with their dates: revenue, cost, headcount, losses avoided or output per employee.

Company	Use case	Return
Alphabet	Gemini tools for ad sales teamsGlobal business organization	Hundreds of millions of dollars of incremental revenue, seller productivity up 10%+ (Oct 2025)Source
Alphabet	AI coding tools and agentsEngineering	Engineering velocity up an estimated 10% from AI, company-wide (Jun 2025)Source
Microsoft	AI translation and partially autonomous supportCustomer Service and Support	About $750 million a year in customer service costs saved, per Judson Althoff (Apr 2026)Source
Microsoft	Supplier sourcing assistantFinance	Estimated $10 million a year in costs saved, from a Microsoft finance slide (Jun 2025)
Microsoft	myDevice AI agentIT operations	Asset-management time and costs down about 50%, early results credited to myDevice AI (Jan 2026)
Salesforce	Help Agent on help.salesforce.comCustomer support	About $100 million taken out of support costs, with satisfaction as good or better (Oct 2025)Source
Salesforce	Agentic codingEngineering	Features and code shipped doubled year over year, engineering headcount flat, credited to AI (May 2026)Source
Amazon	Rufus and Alexa for ShoppingStores	Nearly $12B in incremental annualized sales in 2025, credited to Rufus (Feb 2026)Source
IBM	AskHR and HR agentsHuman resources	Work of a couple hundred HR staff replaced by AI agents, Arvind Krishna said (May 2025)Source
Intuit	AI-assisted and agentic codingEngineering	39% more code delivered per developer, credited to generative AI tools (Dec 2025)Source
Meta Platforms	Agentic coding toolsSoftware engineering	Output per engineer up 30% since the start of 2025, mostly from agentic coding (Jan 2026)Source

Compare

Technology side by side

Company	Ownership	Platform	Models	Hosting	Reach	Stage
Alphabet	CEO Sundar Pichai; Koray Kavukcuoglu, Chief AI Architect, unifies internal coding tools	Built: Gemini, Antigravity and internal coding agents, built in-house	Gemini by default; Claude Opus 5 on a quota inside Antigravity	Google's own data centers, predominantly on its TPUs, plus some leased capacity; Claude hosting for Antigravity not disclosed	~75% of new code AI-generated; 83% of sales staff weekly	Scaling
Microsoft	Kathleen Hogan, EVP Strategy and Transformation; Microsoft Digital runs the rollout	Built: Its own Copilot, Copilot Studio, Foundry and Agent 365	OpenAI by default, plus Anthropic and in-house MAI models	Azure in Microsoft's own data centers plus leased third-party GPUs; Anthropic models run by Anthropic as a subprocessor	Copilot deployed to 300,000+ employees and external staff	Scaling
Salesforce	President Joe Inzerillo (Enterprise and AI Technology), on the executive team; Rohan Kumar runs engineering	Hybrid: Own Agentforce, Slack and Data 360; coding agents and frontier models bought	Anthropic, OpenAI and Google models plus Salesforce's own, chosen by task	Hyperforce on AWS and Google Cloud plus Salesforce-run infrastructure in third-party data centers; Claude via Amazon Bedrock in Salesforce's VPC; Koa on Salesforce's own infrastructure; coding agents vendor-hosted	99% of employees use internal agents (Sep 2025)	Scaling
Accenture	Chief AI and Data Officer Lan Guan, CTO Marc Warner and CIO Tony Leraris, all on the Global Management Committee	Hybrid: AI Refinery and Reinvention.AI built in-house; Copilot and other staff tools bought	OpenAI and Anthropic, also through Microsoft 365 Copilot; Mistral; open-weight Llama and Nemotron	Microsoft 365 Copilot in Microsoft 365 over Accenture's SharePoint and OneDrive data; AI Refinery on NVIDIA AI Foundry, sold as a service on AWS; other internal AI hosting not disclosed	Copilot rolling out to ~743,000; 89% monthly active in a ~200,000-license tranche (Apr 2026)	Piloting
Amazon	CEO Andy Jassy; models and chips under SVP Peter DeSantis; agent tools under AWS	Hybrid: Kiro, Quick, Aza and Nova built in-house; Claude Code and Codex bought	Amazon Nova, Anthropic Claude and OpenAI models, chosen per job	AWS, in Amazon's own data centers, on Trainium, Inferentia and Nvidia GPUs; Claude and OpenAI models served through Bedrock	83% of engineers use Kiro (May 2026)	Piloting
IBM	SVP Transformation and Operations Joanne Wright, under the CFO; CEO chaired the steering committee	Built: Own watsonx stack: Ask agents, AskIBM and Bob	Claude, Mistral and Granite in Bob; others not disclosed	IBM Cloud: Red Hat OpenShift on IBM Cloud VPC, GPU-capable; where Bob's Claude models run is not disclosed	80,000+ on Bob (Apr 2026); AskIT for about 270,000	Piloting
Intuit	Chief AI Officer Ashok Srivastava (Intuit Foresight); CTO Alex Balazs runs technology, data science, IT and security	Hybrid: Own GenOS platform and financial LLMs; coding agents and employee assistants bought	Intuit's own financial LLMs plus OpenAI, Anthropic, Amazon Nova and other commercial models	Mostly AWS (since 2013) and Google Cloud; GenOS built with AWS services including Amazon Bedrock and SageMaker; no Intuit data center named; where OpenAI models and Claude Code run is not disclosed	A majority of employees use AI tools regularly (Dec 2025); no count published	Piloting
Meta Platforms	CTO Andrew Bosworth runs internal AI adoption; Chief AI Officer Alexandr Wang builds the models	Hybrid: MetaCode, Muse Code and Metamate built in-house; Claude Code and Gemini bought	In-house Muse and Llama, plus Anthropic, OpenAI and Google tools	Meta's own data centers (30 owned locations) plus third-party cloud capacity from CoreWeave, Google Cloud and, from 2027, Nebius	Not disclosed; ~30,000 Claude Code and 30,000+ MetaCode users reported (Oct 2026)	Piloting
NVIDIA	Jensen Huang sets the mandate; NVIDIA IT's enterprise AI team runs the internal platform	Hybrid: IT-run AI factory built in-house; coding agents bought	OpenAI, Anthropic and Cursor for staff; open models via NIM; in-house ChipNeMo	In-house AI factory run by NVIDIA IT, NVIDIA's own AI supercomputers, cloud capacity under $27B of multi-year agreements (AWS among the providers, per The Information), and GB200 and GB300 systems serving Codex	10,000+ Codex users across functions (Apr 2026); all software engineers on Cursor	Piloting
Shopify	CEO Tobi Lütke, who also heads R&D; CTO Mikhail Parakhin moving to an advisory role (Oct 2026)	Hybrid: Proxy, Aquifer, River and Quick built in-house; Cursor, Claude Code and Codex bought	Anthropic, OpenAI and Google frontier models plus post-trained open models	Google Cloud for most operations and for the Claude models behind Sidekick; Nebius GPUs for training; Shopify's own GPUs and CentML (NVIDIA) for open models; other frontier models through APIs behind an internal proxy	River touched the work of 7,000+ people in 30 days (May 2026)	Piloting
Apple	SVP Software Engineering Craig Federighi, with VP of AI Amar Subramanya	Hybrid: Enchanté and Enterprise Assistant built in-house; coding tools bought	Apple Foundation Models with Anthropic Claude and Google Gemini, as reported	Apple data centers and private servers; Google Cloud with Nvidia GPUs since Jun 2026; AWS for search inference	Not disclosed; Enchanté reported across office staff (Jan 2026)	Experimenting
Oracle	Global CIO and EVP Jae Evans for employee AI tools; no chief AI officer on the executive list	Hybrid: ChatGPT Enterprise and Codex bought from OpenAI; access controls built on OCI	OpenAI in ChatGPT and Codex; OpenAI and Anthropic in code scanners; Llama in support search	OpenAI-hosted ChatGPT and Codex; Oracle's agents and code scanners run on OCI, including OCI Kubernetes Engine	130,000 active ChatGPT users by OpenAI's count (Oct 2026); none from Oracle	Experimenting
Samsung Electronics	Split by division: DX head Roh Tae-moon and DS head Jun Young-hyun, both co-CEOs	Hybrid: Samsung Gauss and DS Assistant built in-house; ChatGPT, Gemini and Claude bought	Samsung Gauss, OpenAI, Google Gemini, Anthropic Claude; Mistral for chip models	Gauss on Samsung's intranet; Gemini in a dedicated Google Cloud tenant; ChatGPT and Claude on vendor services; chip AI on-premises on Nvidia GPUs	All Korea staff and all DX staff worldwide have access; usage not disclosed	Experimenting

Questions

AI in technology, answered

Which technology companies are furthest along with AI?: In the Enterprise AI diffusion report, Alphabet, Microsoft, and Salesforce are at Scaling, Accenture, Amazon, IBM, Intuit, Meta Platforms, NVIDIA, and Shopify are at Piloting, and Apple, Oracle, and Samsung Electronics are at Experimenting. A stage follows from the returns and rollouts a company has publicly reported in the last 18 months, not from its spending or plans.
How do tech companies use AI internally?: Technology companies use AI internally most in software engineering, customer support, sales, IT and HR. Microsoft credits AI with about $750 million a year of customer service savings (Apr 2026), Salesforce runs its Help Agent on help.salesforce.com and says AI coding tools doubled the features and code it shipped with flat engineering headcount (May 2026), Alphabet says about 75% of Google's new code is AI-generated (Apr 2026) and 83% of its sales team uses Gemini-assisted tools weekly (Jul 2026), and Meta Platforms engineers write code with Meta's own MetaCode and Muse Code while the REA agent runs ads ranking experiments (Mar 2026). Amazon credits its Rufus shopping assistant with nearly $12 billion in incremental annualized sales in 2025 (Feb 2026), says 83% of its engineers use its Kiro coding tool (May 2026), and gives staff internal assistants such as Quick, Aza and MeshClaw. IBM's AskIT agent serves about 270,000 employees, and NVIDIA's IT organization runs an internal AI factory for about 40,000 employees (Jun 2026). Accenture is rolling out Microsoft 365 Copilot to about 743,000 employees (Apr 2026), Shopify's River agent in Slack touched the work of more than 7,000 people in 30 days (May 2026), Intuit runs AI agents in its customer success operations beside human tax and bookkeeping experts, and Samsung Electronics staff use its own Gauss Portal assistant and code.i coding tool.
What return have technology companies reported from AI?: Technology companies report AI returns in a few functions and no firm-level figure. Amazon credits Rufus, its AI shopping assistant, with nearly $12 billion in incremental annualized sales in 2025 (Feb 2026). Microsoft credits AI with about $750 million a year of customer service savings (Apr 2026), an estimated $10 million a year from a supplier sourcing assistant (Jun 2025) and about 50% lower IT asset-management time and costs (Jan 2026). Salesforce credits Agentforce with about $100 million taken out of support costs (Oct 2025) and AI coding tools with doubling the features and code shipped at flat engineering headcount (May 2026). Alphabet credits Gemini tools for its ad sales teams with hundreds of millions of dollars of incremental revenue and over 10% higher seller productivity (Oct 2025), and estimates that AI raised engineering velocity about 10% (Jun 2025). Meta Platforms credits AI, mostly agentic coding, with a 30% rise in output per engineer since the start of 2025 (Jan 2026), and Intuit credits generative AI tools with 39% more code delivered per developer (Dec 2025). IBM credits AI agents with replacing the work of a couple hundred HR staff (May 2025). Accenture, NVIDIA, Shopify, Apple, Oracle and Samsung Electronics report no business return they credit to AI alone.
Do tech companies use their own AI products internally?: Technology companies that make AI mostly run it on themselves first. Microsoft deploys Microsoft 365 Copilot, Copilot Studio, Foundry, GitHub Copilot and Agent 365 under its Customer Zero program, Salesforce runs Agentforce and Slackbot on itself under its own Customer Zero program, IBM builds its internal AI on watsonx, Alphabet runs Google on its own Gemini models, TPUs and Antigravity agent platform, Amazon's engineers and staff use Kiro and Quick, AWS products Amazon also sells, Meta Platforms builds tools such as MetaCode, Muse Code and Metamate on its own Muse and Llama models, and NVIDIA's internal AI factory runs on NVIDIA's own products. Intuit built its own GenOS platform and financial language models, and Samsung Electronics runs its own Samsung Gauss models on its internal network. Oracle bought ChatGPT Enterprise and Codex from OpenAI for its employees (Apr and May 2026), Accenture is rolling out Microsoft 365 Copilot (Apr 2026), and Apple uses its own Foundation Models with Anthropic Claude and Google Gemini, as reported.
Which AI models do tech companies use internally?: Technology companies mostly combine their own models with models from OpenAI, Anthropic and Google. Alphabet uses Gemini by default and has offered Anthropic's Claude Opus 5 to every Google engineer inside Antigravity on a per-user quota since Sep 2026, while Meta Platforms runs its own Muse and Llama models and has steered engineers from Claude Code to its own MetaCode, with Claude Code users falling from about 60,000 to about 30,000 (The Information, Oct 2026). Amazon uses its own Nova models alongside Anthropic's Claude and OpenAI models, and opened Claude Code and Codex to all builders through Bedrock (May 2026). Intuit pairs its own financial language models with OpenAI, under a contract worth more than $100 million (Nov 2025), Anthropic and Amazon Nova, and Samsung Electronics pairs its own Samsung Gauss models with ChatGPT, Gemini Enterprise and Claude for Device eXperience staff (Jun 2026). Salesforce made Claude Code the primary coding agent for all its engineers (May 2026), Microsoft uses OpenAI models by default plus Anthropic and its own MAI models, IBM runs Claude, Mistral and Granite in its Bob coding agent, and Oracle gives staff OpenAI's ChatGPT and Codex (Apr and May 2026).

Other sectors

Every company in the report