Automated machine learning in 2026: overview
Automated machine learning (AutoML) automates parts of the work required to build and optimize machine-learning models, from preparing data and selecting algorithms to tuning parameters. It can reduce repetitive experimentation and make predictive modeling more accessible, but automation does not remove the need for sound data, appropriate evaluation or domain expertise. The practical challenge is to find useful models efficiently while keeping their performance, limits and decision logic understandable.
Recent work is extending AutoML into clinical workflows and more autonomous systems. An October 2026 proof-of-concept used Google Cloud Vertex AI for code-free glaucoma classification from retinal fundus photographs. AutoAIM Studio, introduced in August 2026, brings visual, automated modeling tools to materials science. Meanwhile, research on LLM-driven AutoML agents targets pipeline construction, while meta-learning work addresses the computational cost of evaluating workflows. Together, these directions point toward broader access and more adaptive search, with reliability and deployment constraints still unresolved.
- 1Research broadens: Publications grew 20.2% while patent activity fell 37.8%, suggesting continued research interest alongside slower recent patenting.
- 2Automation gains context: Code-free clinical tools and materials-science applications show AutoML moving beyond general-purpose model building toward workflows shaped by domain needs.
- 3Search remains costly: Meta-learning and LLM-driven agents aim to reduce the time and execution overhead of testing pipelines, but dependable evaluation remains a constraint.
Top organizations working on automated machine learning
Ranked by Linknovate activity score · last 24 monthsThe activity score adds up the relevance of every patent, publication, grant and news item linking an organization to automated machine learning. Showing 7 of 915 organizations.
| # | Organization | Approach | Signals (24 mo) | Momentum | Why it matters |
|---|---|---|---|---|---|
| 1 | 
Microsoft
Corporate · United States |
Automated pipeline search | PUB 6 · NEWS 209 | -68% | Microsoft's Azure Machine Learning automates model and pipeline development, including data preparation, algorithm selection and tuning, for predictive modeling. |
| 2 | 
Amazon
Corporate · United States |
Automated pipeline search | PAT 1 · PUB 6 NEWS 123 |
-82% | Amazon SageMaker Autopilot automates the steps of building machine-learning models, from data preparation and algorithm selection through model evaluation. |
| 3 | 
Google
Corporate · United States |
Automated pipeline search | PAT 2 · NEWS 159 | -69% | Google's Vertex AI AutoML automates model development for tasks such as image, text and tabular-data prediction, making model building accessible through managed cloud tools. |
| 4 | 
IBM
Corporate · United States |
Meta-learning and algorithm selection | PAT 1 · NEWS 155 | -69% | IBM's AutoAI automates data preparation, algorithm selection, feature engineering and hyperparameter optimization to construct machine-learning pipelines. |
| 5 | Automated pipeline search | NEWS 27 | -100% | MyDataModels develops AutoML software that automates model building and helps users generate predictive models from their data. | |
| 6 | 
Oracle
Corporate · United States |
Meta-learning and algorithm selection | PAT 5 · PUB 1 NEWS 59 |
-77% | Oracle Machine Learning automates model selection and tuning, integrating AutoML capabilities with its database and cloud analytics tools. |
| 7 | 
University of Chinese Academy of Sciences
Research · China |
General | PUB 24 | -29% | Researchers apply AutoML to environmental prediction, including evapotranspiration estimation, forest-fire risk and brine salinity modeling. Recently, they used AutoML in a framework for assessing landslide risk along a railway corridor. |
| 8 | ··Hidden organizationCompany | Approach | PAT 00 · PUB 00 GRANTS 0 · NEWS 00 | +00% | Available to registered users. |
| 9 | ··Hidden organizationCompany | Approach | PAT 00 · PUB 00 GRANTS 0 · NEWS 00 | +00% | Available to registered users. |
| 10 | ··Hidden organizationCompany | Approach | PAT 00 · PUB 00 GRANTS 0 · NEWS 00 | +00% | Available to registered users. |
Startups to watch and leading research centers
Startups to watch 6 total
Preferred Networks IncJapanPreferred Networks contributes to black-box optimization research relevant to AutoML and neural-network tuning. Recently, it introduced OptunaHub to bring optimization algorithms and benchmarks together.- ··Hidden startupCountryAvailable to registered users, with funding and founding year.
- ··Hidden startupCountryAvailable to registered users, with funding and founding year.
Leading research institutions 810 total
- University of Chinese Academy of SciencesChinaResearchers apply AutoML to environmental prediction, including evapotranspiration estimation, forest-fire risk and brine salinity modeling. Recently, they used AutoML in a framework for assessing landslide risk along a railway corridor.
Tsinghua UniversityChinaTsinghua researchers apply AutoML to problems such as identifying undiagnosed diabetes and study its role in automated model and pipeline design.
Shanghai JiaoTong UniversityChinaResearchers at Shanghai Jiao Tong University use AutoML for applied risk prediction, including forecasting seaport disruption from natural hazards and assessing postoperative complications.- ··Hidden research institutionCountryAvailable to registered users, with their key researchers.
- ··Hidden research institutionCountryAvailable to registered users, with their key researchers.
Where automated machine learning innovation happens
Share of documents with at least one organization from each country.
Open the interactive map with all 915 organizationsMain technical approaches
Last 24 months| Approach | What it is | Documents | Leading organizations |
|---|---|---|---|
| Automated pipeline search | Searches combinations of data preparation, models and workflow steps; it can find end-to-end pipelines but may consume substantial compute and evaluation time. | 219 |
University of California, Federal University of Pernambuco, University Of Montrealqc |
| Hyperparameter optimization | Tunes model settings to improve performance; it can refine a chosen model efficiently but depends on search budgets and suitable validation. | 271 |
Automated Machine Learning, Microsoft, Shanghai University |
| Neural architecture search | Automates the design of neural-network structures; it can tailor architectures to a task but often requires substantial compute. | 149 |
University of California, Tsinghua University, University of Tsukuba |
| Automated feature engineering | Creates or selects input variables automatically; it can improve predictive signal but may add complexity or produce features that are hard to interpret. | 388 |
Automated Machine Learning, Amazon, Google |
| Meta-learning and algorithm selection | Uses prior task and model-performance information to guide choices; it can narrow search but relies on relevant historical evidence. | 112 |
Leibniz University of Hanover, TU Eindhoven, Sorbonne Polytechnic University |
Emerging trends
Last 12 months vs previous 36 monthsTerms whose frequency in the last 12 months grew the most compared with the previous 36 months.
Funding and investment
Recent funding rounds
Funding data lists 64 grants totaling $93,928,995 since 2020. Separately, Autoscience announced a $14M seed round in March 2026 to automate research and development of machine-learning models.
Patent landscape
Top patent holders · last 24 monthsRecent developments
Latest news, duplicates removed- Oct 2026The development of generative models and AutoML platforms has prompted many specialists and business leaders to reflect on the future of analytics. It appears that algorithms are already capable of writing code…TechBullion
- Sep 2026When building fraud detection systems, standard training steps often hide expensive failures. A model can show 96% accuracy on paper while letting every single fraudulent transaction pass unchecked. In fraud detection…Red Hat
- Sep 2026The best Claude data science tools in 2026 include Claude Data Wrangler Pro for automated data cleaning, the Anthropic Auto-EDA Engine for rapid exploratory analysis, and Claude Python Co-Pilot for advanced scripting.…The Scientific World - Let's have a moment of science
- Aug 2026In this article I will discuss the Best AI Model Hosting Platforms that assist developers, businesses and researchers in deploying, administering and expanding the power of artificial intelligence models easily. - Key…Coinworldstory
- Month 20XXHidden developmentAvailable to registered users with alerts enabled.
Frequently asked questions about automated machine learning
What companies are working on automated machine learning?
Microsoft, Amazon, Google and IBM are among the companies developing AutoML tools. Their offerings automate tasks such as pipeline development, data preparation, model selection, feature engineering and tuning. DataRobot also provides end-to-end predictive modeling, while Oracle integrates model selection and tuning with database and cloud analytics tools.
Which automated machine learning startups are identified in 2026?
Preferred Networks Inc. and Mydatamodels, The Automated Machine Learning Company, are identified as startups associated with AutoML. Preferred Networks contributes to black-box optimization research and introduced OptunaHub, while MyDataModels develops software to automate predictive model building. The available facts do not support ranking these startups by promise or commercial traction.
Which universities and research centers lead automated machine learning research?
The University of Chinese Academy of Sciences, Tsinghua University and Shanghai JiaoTong University are among the research organizations active in AutoML. Their work includes environmental prediction, diabetes identification, automated pipeline design and applied risk prediction. The University of California and Tongji University also appear among active research organizations, with work spanning human-centered tools and zero-code frameworks.
How much funding has automated machine learning received?
Funding data records 64 grants totaling $93,928,995 since 2020. These figures reflect grants classified under the topic, not necessarily funding dedicated exclusively to AutoML. Separately, Autoscience announced a $14M seed round in March 2026 to automate research and development of machine-learning models.
Who holds the most automated machine learning patents?
Tata Consultancy Services Ltd. is the leading named patent holder, with 6 patents; Oracle follows with 5. Align Technology, Prudence Technology Private Limited and 8flow Inc. each have 3 listed patents. These are the holder counts in the topic data, rather than a measure of overall patent quality or commercial value.
Is automated machine learning patent activity growing?
No: patent activity declined 37.8% in the latest 12 months compared with the previous 12 months. The topic data lists 73 patents in its 24-month window. Publication activity moved in the opposite direction, with 1,114 publications in the latest year and growth of 20.2%.
What are the main approaches in automated machine learning?
The five named approaches are automated pipeline search (219 documents), hyperparameter optimization (271), neural architecture search (149), automated feature engineering (388), and meta-learning and algorithm selection (112). They automate different parts of model development, from selecting pipeline components to tuning settings or using prior task knowledge to guide search.
Which countries lead automated machine learning?
The United States leads with 5,009 topic documents (49.3%), followed by China with 1,577 (15.5%) and the United Kingdom with 1,242 (12.2%). India and Germany also have substantial activity, with 1,046 and 1,025 documents respectively. These figures describe document counts, not a ranking of research quality or commercial adoption.
What are the applications of automated machine learning?
Manufacturing (213 documents), finance (159) and healthcare (155) are prominent application areas. Recent clinical work includes code-free glaucoma classification, cardiovascular risk prediction and postoperative risk screening. Other applications in the topic data include retail and commerce and scientific research. Healthcare activity grew 9.5% in the latest 12 months compared with the previous 12 months.
What are the latest developments in automated machine learning?
Recent developments include code-free glaucoma classification using Google Cloud Vertex AI in October 2026 and AutoAIM Studio, a visual AutoML application for materials science, in August 2026. Research also explores LLM-driven AutoML agents and meta-learning to reduce workflow evaluation costs. These directions broaden access and target more efficient search, while reliability and deployment remain challenges.