Meet influential people shaping research, products, infrastructure, education and responsible AI. Then explore a separate top-25 company ranking with profiles, editorial reasons and official source links.
25 profiles
01Helped bring generative AI into everyday use through OpenAI’s products and large-scale deployment.
Profile / organisation ↗首相官邸ホームページ / Office of the Prime Minister of Japan Official Website · CC BY 4.0. Original image and rights information ↗ · Licence ↗
02Combines frontier-model development with work on reliability, interpretability and AI safety.
Profile / organisation ↗TechCrunch · CC BY 2.0. Original image and rights information ↗ · Licence ↗
03Helped build Anthropic’s organisation and commercial operations around its Claude family of AI systems.
Profile / organisation ↗Anthropic — official leadership portrait · Official profile image; rights retained by its owner. Original image and rights information ↗
04Known for combining machine learning and scientific ambition, including the development of AlphaGo and AlphaFold.
Profile / organisation ↗John Sears · CC BY-SA 4.0. Original image and rights information ↗ · Licence ↗
05A central figure in accelerated computing and the GPU infrastructure that supports modern AI training and inference.
Profile / organisation ↗The White House · Public domain. Original image and rights information ↗
06Helped drive Microsoft’s cloud and AI strategy, including AI assistants and enterprise adoption.
Profile / organisation ↗Brian Smale and Microsoft · CC BY-SA 4.0. Original image and rights information ↗ · Licence ↗
07Shapes the deployment of AI across Google’s research, consumer products and cloud ecosystem.
Profile / organisation ↗Photographer: Lukasz Kobus (European Commission) · CC BY 4.0. Original image and rights information ↗ · Licence ↗
08Known for ImageNet and computer-vision research, with a focus on AI that serves people and society.
Profile / organisation ↗ITU Pictures · CC BY 2.0. Original image and rights information ↗ · Licence ↗
09Made foundational contributions to neural networks and deep learning, and is an influential voice on AI risk.
Profile / organisation ↗See the original file page for photographer and image-use details · Image rights described at source. Original image and rights information ↗
10Helped establish modern deep learning and studies how increasingly capable AI systems can be developed safely.
Profile / organisation ↗Xuthoria · CC BY-SA 4.0. Original image and rights information ↗ · Licence ↗
11Known for convolutional neural networks and research into learning useful representations and world models.
Profile / organisation ↗Jérémy Barande · CC BY-SA 2.0. Original image and rights information ↗ · Licence ↗
12Expanded access to machine-learning education and helps organisations turn AI concepts into practical applications.
Profile / organisation ↗TechCrunch · CC BY 2.0. Original image and rights information ↗ · Licence ↗
13An influential deep-learning researcher whose work helped shape large-scale neural networks and language models.
Profile / organisation ↗Eladkarmel · CC BY-SA 4.0. Original image and rights information ↗ · Licence ↗
14A product and engineering leader focused on making advanced AI systems useful and accessible.
Profile / organisation ↗SWinxy · CC BY 4.0. Original image and rights information ↗ · Licence ↗
15Connects AI product development with questions about deployment, governance and the societal effects of powerful systems.
Profile / organisation ↗Christopher Wilson · CC BY-SA 4.0. Original image and rights information ↗ · Licence ↗
16Leads a major semiconductor company supplying processors and accelerators for AI and high-performance computing.
Profile / organisation ↗See the original file page for photographer and image-use details · Image rights described at source. Original image and rights information ↗
17Advances enterprise AI and hybrid-cloud adoption, with a focus on business deployment and governance.
Profile / organisation ↗首相官邸 · CC BY 4.0. Original image and rights information ↗ · Licence ↗
18Helps develop a European AI ecosystem focused on capable models, enterprise use and deployment choice.
Profile / organisation ↗Jan2342342423 · CC0. Original image and rights information ↗ · Licence ↗
19A co-author of the Transformer paper who builds language-model technology for enterprise applications.
Profile / organisation ↗Gabriel Hutchinson · CC BY-SA 4.0. Original image and rights information ↗ · Licence ↗
20Helps grow a shared ecosystem for machine-learning models, datasets, tools and developer collaboration.
Profile / organisation ↗Clément Delangue — Hugging Face profile portrait · Official profile image; rights retained by its owner. Original image and rights information ↗
21Studies AI’s social impacts and promotes research that centres communities affected by automated systems.
Profile / organisation ↗TechCrunch · CC BY 2.0. Original image and rights information ↗ · Licence ↗
22Known for research and advocacy exposing demographic disparities in facial-analysis systems.
Profile / organisation ↗Niccolò Caranti · CC BY-SA 4.0. Original image and rights information ↗ · Licence ↗
23Applies machine learning to biological discovery and drug development, following influential work in AI and education.
Profile / organisation ↗Collision Conf from Toronto, Canada · CC BY 2.0. Original image and rights information ↗ · Licence ↗
24Co-author of a widely used AI textbook and a leading thinker on human-compatible AI systems.
Profile / organisation ↗Bengt Oberger · CC BY-SA 4.0. Original image and rights information ↗ · Licence ↗
25Helped establish the foundations of learning through interaction and reward, central to many modern AI methods.
Profile / organisation ↗Xuthoria · CC BY-SA 4.0. Original image and rights information ↗ · Licence ↗
Explore influential AI businesses across models, cloud, computing, enterprise applications and creative tools, with a profile and an editorial explanation for every position.
25 of 25 company profiles · Original rank numbers stay unchanged when filtering.
AI accelerators, deployment software and developer tooling
NVIDIA supplies GPUs and software used to build and deploy AI systems. Its AI platform includes model deployment tools, agent development frameworks and reference workflows.
Placed first for its enabling role across the AI ecosystem: compute and software support many different model builders and applications. This ranking gives infrastructure leverage greater weight than leadership in one assistant product.
Assess an infrastructure stack for deploying AI applications. This is a suggested use, not a reported client case.
Hardware capacity, operating costs and software compatibility still need workload-specific evaluation.
Research, multimodal models and AI products
Google combines DeepMind research with models and tools spanning language, images, video, robotics and science. Gemini, Gemma and AlphaFold illustrate the range of its AI work.
Second for the breadth connecting foundational research, specialist models and deployable products. It ranks above more specialised labs here because this assessment rewards contributions across several parts of the AI stack.
Explore language assistants, visual generation or scientific AI. This is a suggested use, not a reported client case.
Performance varies by model and task; research demonstrations do not guarantee deployment readiness.
General-purpose assistants and developer models
OpenAI develops general-purpose AI systems and delivers them through ChatGPT and developer services. Its products support writing, research, coding, information analysis and creative work.
Third for connecting general-purpose model development with accessible applications. It sits ahead of enterprise distribution specialists in this ordering because direct model and assistant innovation carries substantial editorial weight.
Draft and review a document or prototype an assistant. This is a suggested use, not a reported client case.
Outputs can be wrong; access, data handling and feature availability depend on the product and plan.
AI distribution through enterprise software and cloud
Microsoft offers AI application and agent development through Azure and integrates AI into workplace and developer tools. Its portfolio connects infrastructure with existing business workflows.
Fourth for the route it provides from AI models into enterprise software and development. Below the leading model-centred entries here, its strongest distinction is deployment and workflow integration.
Add an AI assistant to an existing business workflow. This is a suggested use, not a reported client case.
Implementation quality depends on permissions, data preparation, licensing and integration design.
Claude assistants and task-oriented AI
Anthropic develops Claude for analysis, writing, coding and multi-step work. Claude products support generating drafts and artefacts from user context, with outputs available for human review.
Fifth for a focused combination of general-purpose AI and practical knowledge-work applications. Its narrower infrastructure and consumer distribution footprint keeps it below the broader platforms in this editorial order.
Review a long document and draft an action-oriented summary. This is a suggested use, not a reported client case.
Sensitive work needs suitable data controls and independent checking; fluent reasoning is not proof of correctness.
Consumer AI and model ecosystem contributions
Meta develops AI assistants, models and applications, including Meta AI and Muse. Its AI work also includes the Llama model ecosystem and enterprise initiatives.
Sixth for combining consumer-facing AI with contributions developers can build on. This list rewards that combination, while placing model-and-cloud stacks with broader deployment coverage higher.
Explore a personal AI assistant or a model-based application. This is a suggested use, not a reported client case.
Availability differs by region and product; access to model weights is subject to each release’s licence.
Managed AI services and application infrastructure
AWS offers cloud AI services for building and deploying applications and agents. Amazon Bedrock provides a managed platform within a broader cloud and data infrastructure portfolio.
Seventh for enabling organisations to turn models into operating services. The ordering places direct model and consumer-product influence slightly ahead of its infrastructure-and-deployment emphasis.
Build an AI application using managed models and cloud data. This is a suggested use, not a reported client case.
A broad service catalogue does not remove the need to control costs, model behaviour and data access.
Qwen models and cloud-based AI development
Alibaba Cloud offers Qwen language and multimodal models, APIs and customisation options. Its AI portfolio covers text, coding, audio, vision and applications built around those capabilities.
Eighth for pairing a broad model family with cloud delivery and a model-sharing ecosystem. It leads the next group here through that combination rather than a claim of best performance on every benchmark.
Develop a multilingual assistant or customise a model for a defined task. This is a suggested use, not a reported client case.
Licences, hosting locations and performance differ across model releases and deployment options.
Language models, research releases and APIs
DeepSeek develops language models and provides chat and API access. Its official site links research releases, model documentation and developer services.
Ninth for a focused contribution to the general-purpose model landscape. Its position recognises model development while giving wider cloud, application and infrastructure portfolios additional weight above it.
Evaluate a model for coding or language tasks on a controlled test set. This is a suggested use, not a reported client case.
Deployment decisions require checking release-specific licences, data policies, regional availability and independent task results.
Accelerators and software for AI systems
AMD supplies AI computing products across data centres and devices. Its portfolio includes Instinct accelerators and the ROCm software ecosystem.
Tenth for expanding the hardware and software options available to AI builders. It ranks above application specialists because this methodology values infrastructure choice across multiple downstream use cases.
Assess an alternative compute platform for training or inference. This is a suggested use, not a reported client case.
Software support and throughput must be tested for the specific model; headline hardware specifications are insufficient.
Multimodal research and creative AI services
ByteDance’s Seed organisation develops models for language, visual and audio tasks. Its public portfolio connects research with creative applications and BytePlus developer services.
Eleventh for the range connecting multimodal model work with usable services. This breadth places it ahead of narrower specialists here, without implying that every Seed model outperforms competitors.
Prototype an image, video or audio production workflow. This is a suggested use, not a reported client case.
Product access, content restrictions and rights need review for the intended market and use.
Grok models and developer services
xAI, acquired by SpaceX in February 2026, develops Grok and related AI services. Its public offerings cover language, coding, voice and media through applications and APIs.
Twelfth for its general-purpose and multimodal AI offering. The entry evaluates the xAI business, not SpaceX’s rocket or satellite performance; a narrower AI deployment portfolio limits its position in this cross-sector list.
Evaluate a multimodal assistant or developer API for a defined task. This is a suggested use, not a reported client case.
Model behaviour, sourcing, content controls and permitted uses require independent assessment.
Pangu models for industry and science
Huawei Cloud offers Pangu models spanning language, computer vision, multimodal tasks, prediction and scientific computing. The portfolio emphasises adaptation to enterprise and industry-specific work.
Thirteenth for connecting several AI disciplines to industrial applications. It ranks above narrower software specialists here because of that range, while this assessment does not equate industrial scope with universal model superiority.
Assess a domain-specific model for an industrial prediction task. This is a suggested use, not a reported client case.
Practical suitability depends on deployment access, local requirements, data quality and task validation.
AI models, cloud and autonomous driving
Baidu’s AI stack includes its cloud infrastructure, PaddlePaddle framework, ERNIE models and applications. Apollo Go adds an autonomous-driving business alongside language and enterprise AI.
Fourteenth for a portfolio spanning software foundations and real-world AI applications. It leads the enterprise-software group in this order through that range, while the ranking remains a qualitative comparison.
Study enterprise model deployment or autonomous mobility applications. This is a suggested use, not a reported client case.
Autonomous-driving availability and capabilities are location-specific; they cannot be generalised from a product overview.
Governance, hybrid deployment and business AI
IBM’s watsonx portfolio brings together AI development, enterprise data, assistants and governance. It is designed around integrating AI into existing business infrastructure and controls.
Fifteenth for addressing governance and data integration alongside AI applications. Its position reflects enterprise depth; the methodology gives broader model and infrastructure influence higher priority above it.
Design an AI workflow with access controls and governance requirements. This is a suggested use, not a reported client case.
Governance tooling supports controls but does not, by itself, establish legal compliance or trustworthy outputs.
Data-connected AI applications and evaluation
Databricks provides tools to build AI agents using enterprise data, evaluate outputs and govern access. Its platform connects data management with model and application development.
Sixteenth for linking the data layer to AI evaluation and deployment. It ranks ahead of many narrower applications here because data quality and governance affect numerous downstream AI uses.
Build a domain-specific assistant over governed company data. This is a suggested use, not a reported client case.
A data platform cannot guarantee answer accuracy; retrieval, evaluation and permissions need careful design.
Open-weight models and controlled deployment
Mistral develops general-purpose and specialist models with customisation and deployment options. Its public offering emphasises organisations retaining control over models, data and hosting choices.
Seventeenth for model development combined with deployment flexibility. Its inclusion recognises an alternative route to enterprise AI; larger cross-sector ecosystems receive higher priority in this ordering.
Explore an AI application with defined hosting and customisation requirements. This is a suggested use, not a reported client case.
Open weights do not mean every model or deployment is unrestricted, free or suitable for sensitive work.
Model collaboration, datasets and deployment tooling
Hugging Face operates a platform for sharing and collaborating on models, datasets and AI applications. It also offers inference services, compute and enterprise tools.
Eighteenth for enabling work across many model families rather than one proprietary assistant. That ecosystem role earns a place above several application specialists, even though it is a different kind of AI company.
Explore model cards, datasets and a reproducible AI demo. This is a suggested use, not a reported client case.
A model being hosted on the Hub is not a guarantee of quality, security or permissive licensing.
AI agents within customer and business workflows
Salesforce’s Agentforce platform supports AI agents for customer service and business processes. Its offering connects reasoning, data and actions within enterprise workflows.
Nineteenth for embedding AI into defined business processes. It ranks below general platforms in this list because its strongest contribution is application-layer integration rather than broad foundational infrastructure.
Design a service agent that escalates suitable cases to a person. This is a suggested use, not a reported client case.
Action-taking agents need tested permissions, escalation paths and monitoring; autonomy is not a substitute for controls.
Generative media inside creative workflows
Adobe Firefly supports generating and editing visual and audio content, including access to Adobe and partner models. Its offering connects AI generation with creative production tools.
Twentieth for making generative media part of established creative workflows. The ranking values that application depth while placing companies with broader infrastructure or enterprise-platform roles higher.
Create and refine campaign visuals within a reviewed production process. This is a suggested use, not a reported client case.
Rights, model terms and commercial-use conditions vary; using a creative tool does not remove the need for clearance.
Private enterprise AI and retrieval models
Cohere offers enterprise AI through North and models for generation, embeddings and retrieval optimisation. Its deployment options include private environments for organisations with data-control requirements.
Twenty-first for a focused enterprise stack combining models, retrieval and deployment control. Its narrower consumer and infrastructure scope places it below broader platforms in this cross-sector assessment.
Develop a knowledge assistant using controlled document retrieval. This is a suggested use, not a reported client case.
Private deployment helps control data exposure but does not guarantee correct answers or eliminate security review.
Cloud infrastructure focused on AI workloads
CoreWeave supplies cloud infrastructure and tooling for demanding AI workloads. Its platform connects compute operations with tools for developing and improving models and agents.
Twenty-second for a specialised enabling role in AI computing. It ranks below broader model-and-cloud platforms because its contribution is concentrated on infrastructure rather than a full consumer and enterprise application range.
Assess infrastructure for a demanding model-development or inference workload. This is a suggested use, not a reported client case.
Capacity, contractual commitments, reliability and total operating costs need project-specific diligence.
Web-connected answer and research services
Perplexity offers an AI answer engine and research workflows with source citations. Its platform also includes developer services and browser-based assistance.
Twenty-third for a focused approach to connecting AI answers with web information. Its narrower role places it below foundational and enterprise platforms, while remaining important for everyday research workflows.
Gather initial research and inspect the original sources behind an answer. This is a suggested use, not a reported client case.
Citations do not prove a claim is correct; users still need to read sources and check dates and context.
Speech, audio and conversational agents
ElevenLabs develops AI audio services for speech generation, transcription, dubbing and voice-based applications. Its products serve creative production and conversational enterprise use cases.
Twenty-fourth for specialist depth in audio and voice applications. This position reflects the ranking’s preference for broad ecosystem influence over excellence in a narrower modality.
Produce a reviewed narration or prototype a voice assistant. This is a suggested use, not a reported client case.
Voice cloning requires appropriate consent and rights; language quality and suitability should be tested.
Generative media and world-model research
Runway develops AI models and tools for creative media, developer applications and simulation. Its public offering includes a creative platform, developer services and robotics-related tooling.
Twenty-fifth for combining creative applications with visual and world-model research. It completes a selection that prioritises broad AI ecosystems first; the position is not a judgement that its creative tools are inferior.
Storyboard and iterate a short generated video with human review. This is a suggested use, not a reported client case.
Creative outputs need rights checks and quality review; simulation capability does not establish real-world reliability.
Read the ranking method ↑ · Explore the people shaping AI ↑ · Compare AI models →