data intelligence era: Key trend of enterprise data strategy in 2025-OceanClub technical community.

data intelligence era: Key trend of enterprise data strategy in 2025

GaodeGe! 2025-01-31 23:00 published in China

AI will reshape the storage structure

by 2025, AI will continue to penetrate into various industries, bringing unprecedented opportunities and challenges. The deep integration of AI and storage systems will have a profound impact. AI-driven solutions will become more and more popular, aiming at optimizing performance, strengthening security lines and ensuring data reliability. The increase in AI workloads will create an urgent need for high-performance storage solutions to support data-intensive applications, including LLM, machine learning model training, and real-time data analysis. This will prompt the continuous upgrading of data storage technology to meet the stringent requirements of AI in terms of speed, scalability and efficiency.

-- Boyan Ivanov, CEO of StorPool Storage

storage-level memory and AI Driver optimization will reshape cloud storage efficiency

although hardware innovations such as DNA data storage and quantum storage are still in the laboratory research and development stage, as a technology between SSD and DRAM, it has shown great potential to combine RAM speed with disk storage capacity, although its cost is higher than that of SSD. At the software level, the cloud storage platform will use machine learning algorithms to automatically optimize data storage locations, access modes, and redundant configurations. For example, the AI model is used to predict storage requirements and automatically migrate data between hot storage, warm storage, and cold storage layers to optimize resource utilization.

-- Pat Patterson, chief preacher of Backblaze

enterprises will be more inclined to "moderate principle"

as enterprises prepare for the potential economic recession, cost control will become the top priority, and it will become more and more critical to find the best balance between the storage capacity of key task data and the reduction of storage expenses. Although there are various ways to achieve "moderation" in terms of capacity, performance, scalability, efficiency and manageability, the optimal choice will not only meet current needs, it can also flexibly adapt to future storage systems as workloads evolve.

-- Judy Kaldenberg, senior vice president of sales and marketing, Nexsan

the trend of data localization will accelerate.

In 2025, data localization will become an important trend. Although cloud storage is still crucial, stricter regulations and AI-driven storage demand growth will prompt people to pay more attention to the idea of storing data near the collection site. At the same time, data storage will be closer to each other. Edge Computing scenarios, especially for emerging technology fields such as self-driving cars, smart homes, and AI models that require real-time and real-world data to run quickly and avoid interruptions.

-- Sterling Wilson, on-site chief technology officer of Object First

implement energy-efficient storage while meeting AI requirements

with the growth of AI workloads, its energy demand and related costs continue to rise, prompting enterprises to seek strategies to save costs and improve energy efficiency. Data storage will become the core focus, because large datasets are critical to AI, but their maintenance costs are high. Enterprises will increasingly tend to adopt scalable, low-power storage solutions and use cold storage to store infrequently accessed data to reduce energy consumption. However, these "cold data" will not be idle, but will be actively restored for reuse, re-realization and re-calibration of the model according to business requirements. By balancing high-performance access with efficient cold storage, enterprises can meet AI requirements while reducing costs and environmental impact.

-- Tim Sherbak, product and solution manager, Quantum Enterprise

enterprises will pursue all-purpose storage systems

in order to gain flexibility to accurately match storage requirements, enterprises will seek systems that integrate multiple functions, such as hybrid arrays that combine Flash performance with deep storage capacity of mechanical hard disks (HDD). A storage system that can efficiently integrate different platforms and protocols (such as block storage and file storage) to support multiple application scenarios and provide diversified data management options to ensure data protection, security, and business continuity, it will be most favored by enterprises that need to balance financial and operational needs.

-- Judy Kaldenberg, senior vice president of sales and marketing, Nexsan

future-oriented data requirement layout

there is a popular proverb: The best time to plant a tree is ten years ago, followed by now. In 2025, while adjusting the storage infrastructure to adapt to the current business environment, enterprises need to plan in a forward-looking manner to cope with the inevitable data growth. No matter what storage protocol or hardware is used, it must be able to flexibly adapt to changing workloads, data types, security and compliance policies. In order to better meet current and future storage requirements, more enterprises will adopt scalable storage systems in 2025. These systems can expand with the increase of capacity requirements, provides greater flexibility for future needs.

-- Judy Kaldenberg, senior vice president of sales and marketing, Nexsan

software defined storage (SDS) becomes the first choice for storage

as enterprises are increasingly turning to hybrid cloud and multi-cloud environments, traditional hardware storage systems can no longer meet the needs of modern IT infrastructure for agility, scalability, manageability and cost-effectiveness. By running on standard servers and networks, SDS decouples data from underlying hardware, providing unparalleled flexibility between local data centers and cloud environments to deploy, manage, and expand storage resources. In many hardware upgrades and data center integration projects, SDS is becoming the preferred solution. In the future, we will continue to witness the transformation to a fully automated, software-defined storage solution, focusing on performance, security, manageability, API, and cost reduction in larger and more complex infrastructure projects.

-- Boyan Ivanov, CEO of StorPool Storage

AWS S3 will become the preferred storage for large-scale applications

by 2025, more solutions specially built around AWS S3 will emerge. In the past, developers usually used S3 for cold storage, because the storage cost of S3 is relatively low, but the access cost is relatively high. However, over time, developers have overcome the restrictions on access costs through caching and other technologies. These technologies have been widely adopted, making S3 the preferred primary storage for large-scale applications. Developers can take advantage of the high reliability and availability of S3 without worrying too much about costs.

-- PingCAP software architect Sunny Bains

network Storage becomes an active defense tool

by 2025, the increasingly severe network threat environment will make Cyberstorage a key feature of enterprise storage solutions. Network Storage integrates advanced security measures directly into storage systems, such as AI-driven threat detection, automated response, and isolated Immutable Backups. These functions transform storage from passive assets to active defenders of network attacks, providing real-time protection for enterprises to prevent data leakage, ransomware attacks, and other malicious activities.

-- CTERA chief technology officer Aron Brand

the rise of unstructured data Lake

in 2025, the booming development of enterprise AI agents will push enterprises to build an unstructured data lake oriented to AI. These data lakes will be designed to process diversified unstructured data for retrieval of enhanced generation (RAG) and large language models, ensuring that data collection, storage, and collation can optimize machine learning models. With strong security and compliance control, these AI-ready data lakes will help enterprises explore new insights from data and promote AI to play a central role in decision-making and innovation, covers all business functions of the enterprise.

-- CTERA chief technology officer Aron Brand

immutable storage will be standard

with the increasing frequency of ransomware attacks and the improvement of compliance requirements, the demand for WORM immutable storage will become common by 2025. The storage system is designed to create an Air-Gapped and tamper-proof repository. Even if cyber criminals obtain administrator privileges, they cannot delete or tamper with data. This standardization of immutable storage ensures that enterprises can always restore clean data copies, provides stronger protection than traditional backup repositories, and becomes an important defense against malicious encryption or data damage.

-- CTERA chief technology officer Aron Brand

the rise of hybrid Lakehouse model

the revival of the local data architecture will promote the expansion of Lakehouse into a hybrid environment to achieve seamless integration between the cloud and local data storage. The hybrid Lakehouse model not only has the scalability of cloud storage, but also has the security control capability of local storage, providing flexibility and scalability in a unified and accessible framework.

-- Starburst co-founder and CEO Justin Borgman

SQL recovery in data Lake

with Apache Iceberg and other tabular formats simplifying data access, SQL is gaining popularity in the data lake, making the SQL engine outperform Spark. The revival of SQL has made data more democratic within the organization, promoted data-based decision-making and improved the team's data literacy. The ease of use of SQL will make data insight within easy reach, thus promoting the implementation of data empowerment.

-- Starburst co-founder and CEO Justin Borgman

modern data-driven SaaS applications will be built based on data Lake

new data applications will be based on data lakes rather than non-traditional databases or data warehouses. The reason for this trend is simple: SaaS companies attach great importance to the gross profit margin of their products, while data Lake has significant advantages in total cost of ownership (TCO) and there is no risk of supplier locking. Applications built based on the object storage lake can be stored in open formats such as Iceberg and open computing engines such as Trino. Finally, a cost-effective application stack that can support the scale of the Internet will be built.

-- Starburst co-founder and CEO Justin Borgman

global data explosion may lead to storage shortage crisis

global data generation is growing at an unprecedented rate. It is predicted that by 2028, the global data volume will reach 400 ZB, with an annual compound growth rate (CAGR) of 24%. As a comparison, research by California Institute of Technology shows that only 1 ZB of data volume is equivalent to the amount of information of all sand grains on the Earth. With the maturity and scale of AI, the value of data will be further enhanced, prompting us to store more data and prolong its retention time. However, the annual compound growth rate of storage infrastructure was 17%, significantly lower than the growth rate of data generation. In addition, it takes a whole year to make a hard disk. This gap in growth rate will break the global balance of storage supply and demand. As enterprises change from experimental to strategic in AI applications, they need to formulate long-term capacity plans to ensure storage supply and fully realize the return on AI infrastructure investment.

-- Seagate Technology Executive Vice President and Chief Business Officer B .beautteh

storage Innovation is the key to coping with data center pressure and protecting the environment

as the data explosion continues, the data center may eventually be overwhelmed. However, financial, regulatory, and environmental constraints will increasingly pose challenges to the need to expand the space and capacity of physical data centers. For example, the British State Grid predicts that the demand for electricity in commercial data will increase six times in the next 10 years alone. In addition, CBRE predicts that the progress of AI will significantly promote the demand for data centers in the future, and high-performance computing will require fast and innovative data center designs and technologies to cope with the increasing demand for power density. However, solving these problems requires not only innovation in computing. Hard disks with higher Areal Density can increase the data capacity per unit of storage medium, thus expanding the data capacity without creating a new data center, significantly save TCO and reduce the impact on the environment.

-- Seagate Technology Executive Vice President and Chief Business Officer B .beautteh

2025 will be the first year of AI proxy

This new AI application will not be limited to generating text or images, but will be able to perform operations. For example, they can study topics on the network, manipulate applications on the PC desktop, or complete other tasks through APIs. Although there is still a long way to go from general AI(AGI), these early agents will have highly specialized functions. We will witness the emergence of "Agentic Architectures", that is, focusing on application scenarios that can immediately create value. Possible examples include data modeling, master data management, analysis, and data enhancement. These tasks are highly structured and existing prototypes show potential. The first batch of case studies will emerge in 2025, and then enterprises will follow up quickly, especially after witnessing competitors gain advantages due to AI agents.

-- Weaviate CEO Bob van Luijt

data management will drive business and decision intelligence

in 2025, the core goal of data management is to promote business intelligence and decision intelligence. To achieve this goal, you need to build an Insight Foundation throughout the application lifecycle and integrate security, governance, privacy, data persistence, and compliance. Enterprises will shift from passive Data collection to the creation of standardized and operable intelligence, and realize the interconnection of cross-interconnection systems through technologies such as Data Fabric and Data Mesh. This transformation will make traditional infrastructure face severe challenges because IT cannot support cross-system integration and real-time analysis, prompting IT leaders to give priority to improving ecosystems. This change has upgraded data literacy from technical skills to organizational capabilities, and promoted IT teams to cultivate data awareness in various functional departments. At the same time, AI applications will be accelerated to mine data insights, generate decision intelligence, and realize intelligent automation.

-- Vertex Chief Technology Officer Sal Visca

enterprises that reshape data into dynamic assets will flourish

by 2025, the enterprises that can succeed will be those organizations that reshape data into dynamic assets. Data will no longer be static resources, but the core driving force for innovation, decision-making, productivity, continuous compliance and control, and operational resilience.

-- Vertex Chief Technology Officer Sal Visca

GPU-centered data orchestration becomes a top priority

as 2025 approaches, a core challenge in AI/ML architecture is still how to efficiently transmit data between GPU and GPU, especially remote GPU. As enterprises expand their AI/ML workloads in distributed systems, GPU access is becoming a key architectural issue. Although traditional data orchestration solutions are still valuable, they are increasingly unable to cope with the demand of GPU-accelerated computing. The bottleneck lies not only in data flow management, but also in how to optimize data transmission to GPU, especially remote GPU, to support high-performance computing (HPC) and advanced AI models. Therefore, the innovation of data orchestration solutions centered on GPU will usher in a wave. These new systems will minimize latency, increase bandwidth, and ensure seamless data transmission between local and remote GPUs.

Enterprises have realized the importance of this problem and are trying to rethink how to process data pipelines in GPU-based architectures. It can be predicted that enterprises will increase their investment in technology to simplify data transmission, give priority to improving hardware efficiency, and implement scalable AI models, therefore, it is booming in a distributed and GPU-driven environment.

-- Jasmine Presley, senior vice president of global marketing, Hammerspace

eliminating data islands will become the core task of AI and data architects.

By 2025, eliminating data islands will become a key architectural concern for data engineers and AI architects. Integrating and unifying different datasets within an organization will be a necessary condition for promoting advanced analysis, AI, and machine learning programs. With the increasing number and diversity of data sources, overcoming these isolated islands is crucial to realizing the overall insight and decision-making capabilities required by modern AI systems. The focus will shift from infrastructure to seamless data integration across platforms, teams, and regions. The goal is to create an ecosystem that makes data easy to access, share, and operate in various fields. New tools and frameworks are expected to emerge to simplify data integration and facilitate greater collaboration in traditional isolated island environments.

-- Jasmine Presley, senior vice president of global marketing, Hammerspace

enterprise HPC needs standardized technology integration with unstructured data processing

by 2025, medium and large enterprises will face a key challenge: how to adopt high-performance computing (HPC) in unstructured data processing and meet enterprise standards. As enterprises increasingly rely on AI and data analysis to gain competitive advantages, the need to process large amounts of unstructured data (such as text, images, and videos) is inevitable. However, due to the complexity of coordinating dedicated HPC technologies with enterprise security, compliance, and operational standards, enterprises have long faced difficulties in adopting HPC on a large scale.

The solution is to develop HPC technologies that can run in Enterprise Standard environments. It is expected that enterprise HPC solutions will emerge by 2025, which can seamlessly integrate with standard clients, operating systems, networks, and security frameworks. This integration will enable enterprises to make full use of HPC in large-scale unstructured data processing without sacrificing their security, compliance, or performance standards.

-- Jasmine Presley, senior vice president of global marketing, Hammerspace

2025: The Rise of Collaborative Global Namespace

in 2025, enterprises will redefine the data processing policies in the industry for the management of Global namespaces. Not all global namespaces have the same function: some only support read-only operations, while others can actively read and write. Although Unified View sounds quite efficient, its real core value lies in its seamless operation of these data. If a team cannot collaborate on the same dataset in real time and needs to create multiple replicas for complex merging, the goal of simplifying data management is impossible. The problem of data replicas flooding is that multiple users create different versions of datasets for their respective read and write tasks, which may lead to inefficiency, data Islands, and data inconsistency.

As enterprises are committed to building more collaborative and efficient data environments, they need to preferentially deploy global namespaces that support active read and write. This can not only effectively avoid data fragmentation, but also achieve seamless collaboration, making the enterprise's data infrastructure elegant and efficient under modern workloads.

-- Jasmine Presley, senior vice president of global marketing, Hammerspace

Synthetic Data will become the main driving force for AI to drive consumer insight

with the powerful Foundation Model, synthetic data can solve two core problems: one is to obtain sufficient data to train AI models, and the other is to avoid using customer personal data. As more and more companies adopt synthetic data, its accuracy will continue to improve until AI-driven consumer insight tools widely win trust between brands and consumers.

-- Mike Diolosa, chief technology officer of Qloo

strategic use of synthetic data will become an advantage rather than an obstacle

as more and more organizations explore the great potential of synthetic data, the cognition of this technology will inevitably change. Synthetic data is a form of data that is statistically highly consistent with real-world data and does not need to rely on third-party data collected or purchased manually. It will become an important strategic advantage to extend the generation of synthetic data to many industries including medical and manufacturing. In the future, the possibility of using this data is almost unlimited.

-- Susan Haller, senior analytical director of SAS

data problems intensify AI differentiation

by 2025, some enterprises will stand out in the field of GenAI, surpass competitors, create highly personalized customer experience, and accelerate the launch of innovative products. However, some enterprises are left behind in the competition of generative AI. They may give up projects that began in 2023 because they ignore a key fact: AI needs high-quality data. Poor data can hinder AI performance, and enterprises need to have the courage to trace back and solve widespread data problems.

-- Marinela Profi, head of global generative AI strategy for SAS

end of traditional BI: API priority and generative AI embed analysis into each application

by 2025, traditional BI tools will be removed from the historical stage and replaced by API-First Architecture and generative AI, which seamlessly embed real-time analysis into each application. Data Insight will flow directly into CRM, productivity platforms and customer tools, enabling employees at all levels to make data-driven decisions immediately without technical expertise. Companies embracing this transformation will release unprecedented productivity and customer experience, while static dashboards and isolated systems will be eliminated.

-- Ariel Katz, CEO of Sisense

data literacy will become a mass movement-driven by Composable Apps

by 2025, the mass data literacy movement promoted by combinable applications will fully emerge. These applications seamlessly integrate real-time analysis into daily experience. Consumers can actively understand data on energy use, shopping habits and sustainability through an intuitive and easy-to-use platform. Companies that simplify data reporting and empower users will succeed, while companies that rely on obscure and complex reports may face a strong rebound in consumer demand for transparency.

-- Ariel Katz, CEO of Sisense

AI becomes a tool for improving analysis team skills

in the coming year, AI will completely reshape the analysis team so that non-technical members can also be competent for advanced analysis tasks. AI-driven technical methods will break the traditional technical barriers. From product managers to end users, everyone can actively participate in complex data projects. This change will not only enhance productivity, but also promote a culture of seamless collaboration and accelerate the innovation process within the organization.

-- Yigal eaping, senior vice president, Sisense products and strategy

the boundary between no code and professional code analysis will be broken

by 2025, the boundary between No-Code and Pro-Code analysis will No longer exist. The AI-driven platform will enable product managers to generate 80% of the analysis content and allow developers to optimize and adjust it. This revolutionary analysis and development method will significantly shorten the development cycle and maximize team efficiency, thus thoroughly reshaping the integration method of Enterprise analysis and making data-driven decisions more collaborative and accessible.

-- Yigal eaping, senior vice president, Sisense products and strategy

self-service analysis will become the new normal for end users

as user empowerment needs grow, Self-Serve Analytics will become a trend. Enterprises will increasingly allow customers to build and customize their own dashboards, making data more accessible and valuable at every level of the user experience.

-- Ronen Rubinfeld, senior vice president of Sisense engineering

AI improves data quality

AI will refocus on improving data quality for two main reasons: first, high-quality data is the basis for training and fine-tuning models; Second, AI-driven analysis tools will provide higher-resolution data views to reveal previously undiscovered data quality problems.

-- Ryan Janssen, CEO of Zenlytic

real-time data observability will become the key

Real-Time Data Observability will become crucial. Enterprises need to ensure Real-Time visualization of dynamic Data streams. With in-depth insight into dynamic data workflows, the team can optimize the data pipeline in real time, significantly improving the system response capability and overall operational efficiency. Data pipelines can not only run smoothly, but also evolve as business requirements change.

-- Somesh Saxena, CEO and founder of Pantomath

data observability will become an important trend in 2025

Correctly implemented Data Observability will become an important tool for enterprises to maintain Data correctness. Combining data with the observability of AI is crucial for enterprises that want to make full use of AI. Observability will contribute to security and governance, enabling enterprises to always stay ahead in data persistence, ETL processing, applications, BI reports, or ML/AI pipelines. However, observability needs to be active. For example, it is not enough to simply understand the decline in data freshness and see this in static display. Observability requires intelligent automation or notification of relevant personnel to take action to trigger a response.

-- Kunju Kashalikar, senior director of Pentaho product management

enterprises will adopt advanced data management strategies to promote AI production

enterprises will begin to deploy better data management strategies to classify, tier and store data according to value and purpose to support AI and GenAI production objectives. Relayering and automated storage based on data value and usage will reduce infrastructure and data management costs, enabling teams to focus more time on high value-added tasks while saving budgets, help AI and generative AI move from the pilot phase to large-scale cost-effective production. Strong data classification will also bring significant advantages in processing personal identity information (PII), confidential information, and avoiding deviations when training models.

-- Kunju Kashalikar, senior director of Pentaho product management

data governance will become the top priority in the AI era

enterprises have basically determined the selection of Data repositories (such as Snowflake and DataBricks). The next focus will shift from simple AI experiments to comprehensive Data Governance. This means giving priority to access control, synchronization, translation, documentation, and problem solving, not just experiments. Enterprises are preparing for the AI revolution by centralizing data into these repositories. However, many enterprises are still running isolated systems, with copies of relevant data scattered everywhere and interpreted through their own models. This not only increases the threat surface, but also increases the demand for transparency and control of data access, especially in the era of frequent data leakage and privacy fines. Although 88 percent of data owners admit that data security will become the top priority in 2025-even beyond AI-many CEOs still put growth first. This is a balance, but as enterprises adopt the Shift-Left Approach for data governance, they will give priority to strengthening data governance and security. This means extending from the cloud data warehouse to the source system of data streams and building strong governance capabilities.

-- Jonathan Wright,MetaRouter senior sales engineer

personalization will become the primary role of AI in the growth of digital wallet

digital wallet is completely changing the traditional application experience by providing personalized configuration files for transactions, preferences, and personal data. Wallet technology is used to simplify payment, identity and credential management, etc. With the advent of AI, these two technologies will be combined to provide ultra-personalized experience, prevent fraud and gain insight. GenAI has become a hot topic in the past two years, and AI personalization will become the theme in the next two years. Wallet technology for managing personal data will become the core pillar of these measures.

-- Oz Olivo, vice president of Inrupt product management

digital wallet will go beyond voucher and card management

the personal data wallet that complies with privacy regulations is expected to achieve innovation with the help of Web 3.0 protocol in 2025. Its function will go beyond simple credential management and evolve into a life prediction navigation system. These systems will integrate data from wearable devices, smart home sensors, social networks and environmental monitors to provide users with important insights into future health conditions. They can capture subtle patterns, such as gait changes within six months, vitamin D exposure, sleep quality, and bone mineral density trends, thus giving early warnings several months before potential health problems occur, problems were found earlier than traditional diagnostic methods. Similarly, these wallets will also strengthen personal cognition on career network interaction, skill development mode and industry trend, and help individuals identify career opportunities or risks that are difficult to detect by themselves, just as the innovation of glasses technology has greatly improved human intelligence. More importantly, these systems can not only display data, but also understand the complex interactions between personal bodies, social contacts and professional life, and provide comprehensive suggestions accordingly. For example, they may suggest changing office locations to reduce the impact of air pollution on cognitive performance, or point out how children's recent sleep disorders are associated with declining test scores and changes in family evening habits.

-- Davi Ottenheimer, vice president of trust and digital ethics, Inrupt

AI will drive the next generation of QR code analysis

AI integration in the QR code platform will focus on data analysis and insight generation, rather than the creation of QR codes. The platform will deploy AI as an auxiliary tool to analyze a large amount of behavior data collected through two-dimensional code interaction, helping enterprises understand the scanning mode, frequency and customer travel flow across multiple locations. This function will enable enterprises to process two-dimensional code data on a large scale and extract actionable insights without manually analyzing weekly or monthly reports.

-- Uniqode co-founder and CTO Ravi Pratap

2025: Real-Time RAG leads a new era of dynamic insight

as organizations break the limits of batch processing, real-time retrieval enhanced generation (RAG) technology will emerge. The current RAG implementation mainly relies on the static large language model (LLMs) and batch vector database the response is enhanced by preprocessing static data. Although this method is effective in many applications, it does not perform well in dynamic scenarios requiring real-time information update, such as logistics optimization, personalized video game assistant or financial risk monitoring. Real-Time RAG will make up for this deficiency by combining LLMs with real-time data streams and event-driven architectures, so that the model can access and utilize the latest data during generation. This change will release powerful real-time insights in scenarios that require real-time context, thus making 2025 a key year for real-time enhanced intelligence development.

-- StarTree co-founder and CEO Kishore Gopalakrishna

from data flow to insight: 2025 marks the arrival of the real-time analysis revolution

real-time analysis will usher in a major leap in 2025, as enterprises will complete the "last mile" of the data architecture ". In the past few years, enterprises have attached great importance to construction. Apache Kafka such as event flow system to ensure real-time data flow. However, many enterprises have now realized that traditional analysis terminals (such as data warehouses and batch processing-based solutions) cannot fully realize the potential of these data streams. These traditional systems cannot provide the immediate insight required by today's fast-paced environment. By 2025, enterprises will give priority to real-time analysis platforms that can process, analyze, and respond to data in real time to open the closed loop of data streams and release the real value of the streaming architecture. This change will promote the development of innovative use cases such as ultra-personalized customer experience, real-time external-oriented data products, and adaptive risk management systems. Its capabilities are far beyond those of traditional solutions.

-- StarTree co-founder and CEO Kishore Gopalakrishna

knowledge Map: to make data more intelligent and connect generative AI with users

knowledge Graph provides a semantic layer that can describe the enterprise data ecosystem in human terms and establish new logical connections between previously dispersed data sources. As the GenAI model is considered in a more humanized way, the knowledge graph enables the model and business users to "understand" the available data and generate real insights based on it.

-- Christian Buckner, senior vice president of Altair analysis and IoT

knowledge Graph: innovating data interaction providers type

as enterprises seek to gain competitive advantages through data democratization, knowledge maps provide a more convenient way for business users to access and use data. The Knowledge Graph is like an intelligent assistant that organizes scattered data into a format that can be easily understood and used by human beings and AI. This not only simplifies the ability of generative AI to provide valuable insights, but also enables people to make smarter decisions, just as virtual assistants help plan road trips.

-- Christian Buckner, senior vice president of Altair analysis and IoT

Agentic AI: a new force to change data analysis

nowadays, many business leaders are in trouble not knowing what questions to ask data or where to find answers. The AI agent (Agentic AI) is changing this situation. It can automatically provide insights and suggestions without any initiative to initiate inquiries. This level of automation is crucial to help enterprises deeply understand data and explore its links, thus enabling them to make more strategic business decisions. Enterprises need to establish protection mechanisms to control AI-driven recommendations and maintain trust in results.

-- Christian Buckner, senior vice president of Altair analysis and IoT

dialogue AI: a new tool for business transformation

by 2025, dialogue AI will become one of the most important practical applications of generative AI, greatly promoting the growth of self-service analysis in Banking, Financial Services and Insurance (BFSI) industries. Enterprises will increasingly deploy this technology internally to enable employees to quickly interpret enterprise-wide data, thus enhancing decision-making ability and improving transaction efficiency, and completely change the way enterprises use data resources to obtain competitive advantages.

-- Dylan Tancill, head of global BFSI in Altair

the regulatory demand for self-help analysis tools in the financial field is increasingly prominent.

Self-Service analysis and End User Computing (EUC) tools are popular, many of which are developed internally, it will promote the establishment of a more sound governance framework and protective measures within the enterprise and even the entire financial industry. As regulators pay more and more attention to risk management, enterprises, especially banks, will begin to identify high-risk use cases related to these tools. To cope with this trend, they will implement structured supervision to clarify the scope of EUC and available tools. This will prompt enterprises to conduct a comprehensive evaluation of existing tool sets to ensure compliance and reduce potential risks.

-- Dylan Tancill, head of global BFSI in Altair

low code/no code tool: a solution for resource-intensive manufacturers

the manufacturing industry will continue to face major challenges such as supply chain problems, labor shortage and fierce global competition. In this environment, improving efficiency through technology will become crucial. Due to limited budgets, manufacturers will be more cautious in their investment in technology. Low-Code and no-code AI tools will provide engineers with significant value, enabling them to utilize data analysis without mastering complex coding skills. This shift will help manufacturers achieve innovation and process optimization without increasing the recruitment costs of data scientists, thus maintaining agility and competitiveness with limited resources.

-- Scott Genzer, Altair data science expert

relying on manual tagging as the core teaching Source of AI training

training AI is similar to educating young children. Although young children are talented and can quickly absorb new knowledge, they are also very confused. The input AI information is integrated and parsed, and the AI response is based on these inputs. If the data is mixed with error information, harmful content or prejudice, the response may also map these problems. In general, this poses a challenge to data annotation and basic fact definition. Because of this, the team should rely on manual labeling as the core teaching source, rather than relying on general data sources such as social media. Looking forward to the future, solutions in AI training will be more refined. When data is labeled, facts, speculations, theories, assumptions, and other types will be distinguished.

-- Michael Armstrong, chief technology officer of Authenticx

chief Data Officer (CDO) to reshape data strategy in 2025

The rapid development of AI opens up new opportunities for advanced analysis and helps enterprises dig deep into the value of existing data. In view of the increasingly complex global data environment, enterprises need to adjust data fusion and analysis methods to achieve strategic goals, which is especially critical. Making good use of available data can not only reveal new insights, but also drive operational efficiency, create new market opportunities, and meet employee training needs. However, the potential of any strategy is based on a robust data environment. Entering 2025, CDO must inject new value into its role by ensuring that enterprises make full use of accessible data. Although CDO positions traditionally focus on compliance and risk management, they must present the potential opportunities brought by data analysis to the leadership while performing traditional data management responsibilities. There is a significant gap between CDO's compliance defense and the promotion of business and mission achievements. In the new year, efforts should be made to bridge this gap.

-- Nightwing chief technology officer and chief data officer Chris Jones

organizations turn to a unified platform to protect and manage data

in 2025, with the further adoption of cloud platforms and the wide application of AI, data management will become the focus. In the context where data is intentionally or unintentionally accessed, it becomes crucial to ensure that the right people and models can access the data. Data security used to rely on the combination of single-point products such as Discovery, classification, marking, and loss prevention. However, the current market is tending to be a platform that can simultaneously provide multiple functions. The accelerated development of the unified platform will meet diversified needs, but many organizations will still face the challenge of how to properly classify and protect data, not only to prevent threatening actors, it is also necessary to avoid unexpected access caused by internal goodwill actors.

-- Justin Flynn, senior director of Stratascale

overcoming Data access challenges becomes a core element of AI success

in 2025, with the increasing demand for AI workloads and the trend of distribution, enterprises will face greater pressure and urgently need to solve the challenges of data access. The explosive growth of data in multiple clouds, regions, and storage systems has led to significant bottlenecks in data availability and mobility, especially in compute-intensive AI training. Enterprises need to efficiently manage data access in a distributed environment while minimizing data movement and duplication. We will witness the development of technology to quickly and concurrently access data, regardless of where the data is, while maintaining data localization to optimize performance. Overcoming these data access challenges will become a key differentiation factor for enterprises to expand their AI programs.

-- Li Haoyuan, founder and CEO of Alluxio

AI-driven security analysis for real-time failover

as AI-driven security analysis tools become standard for threat detection and response, enterprises will give priority to ensuring high availability of these applications to achieve uninterrupted operation. Failover clusters will play a key role in maintaining continuous real-time threat detection and response to prevent vulnerabilities in security coverage from putting enterprises at risk. With Failover Clusters, enterprises can reduce risks and ensure the continuous operation of key security monitoring and analysis tools.

-- Cassius Lu, vice president of customer experience, SIOS Technology

cloud-native solutions shape the future of data security

as data is distributed in diversified cloud-native architectures, adaptive and data-centric security is crucial. Cloud-native solutions now provide dynamic protection during the data lifecycle to ensure data security in static, transmission, and use. In 2025, with stricter compliance standards and increased data-centric attacks, cloud-native solutions will play a key role in maintaining resilience, adapting to new regulations, and responding to changing threat environments.

-- Moshe Weis, chief information Security officer of Aqua Security

energy availability will become a core constraint on the growth of quantum computing, AI and data analysis.

As the demand for quantum computing, AI and large-scale data analysis continues to grow, the industry will face a major obstacle: energy shortage. By 2030, data centers may consume as much as 10% of the world's energy, while many key areas (such as Virginia and Texas) are approaching the capacity limit. Illinois is one of the few areas where energy is still available, but its supply is also rapidly consuming. In the future, the development of these technologies will depend on obtaining a large amount of clean energy, giving priority to energy efficiency by enterprises and adopting sustainable energy to maintain competitiveness.

-- Chris Gladwin, CEO and founder of Ocient

cloud analysis is becoming a secondary role, and enterprises are turning to cost-controllable and predictable solutions.

Enterprises are about to see a major change in data analysis methods. Cloud solutions are questioned because of the opaque billing methods of suppliers. These methods often lead to unexpected expenditures and undermine financial planning. At present, more than half of the companies regard cloud spending as the main problem, but lack the necessary visibility to truly control or optimize these costs. Without understanding the actual usage and application requirements, enterprises are like "blind driving", similar to driving without fuel consumption meters. In the next 12 to 18 months, this lack of transparency may push enterprises to turn to more predictable and controlled hybrid models and alternatives on a large scale. Before cloud providers provide the necessary transparency to support accurate spending supervision, the cloud model will take a secondary position and enterprises will seek sustainable and controllable solutions, to effectively manage and optimize its usage.

-- Chris Gladwin, CEO and founder of Ocient

enterprises must refocus on data

although generative AI has attracted wide attention, most enterprises have not yet solved the prerequisite for unlocking its value-underlying data. The large language model (LLMs) on which generative AI depends depends depends entirely on the quality of enterprise data. It is not difficult to find that the generating AI projects of many enterprises are difficult to break through the pilot stage, because of the lack of modern data strategies. In order to realize the return on generative AI investment in 2025, enterprises are expected to focus on technologies that can effectively collect and manage large amounts of unstructured data.

-- Chief Cloud Strategist of Pluralsight Drew Firment

retaining large data sets is a must.

Generative AI relies on a wide range of structured, unstructured, internal, and external data. Its potential is based on a powerful data ecosystem that supports training, fine tuning, and retrieval of enhanced generation (RAG). For industry-specific models, enterprises must keep large amounts of data for a long time. With the changes of the world, relevant data often show their value only after the event, revealing the inefficiency and opportunities. By retaining historical data and combining it with real-time insight, enterprises can transform AI from experimental tools to strategic assets to promote practical value throughout the organization.

-- Lenley Hensarling, technical consultant of qianpike

instant Data Insight

enterprises will give priority to real-time analysis and provide insights within a few minutes to meet growing customer and market demands and competitive pressures. This change will enable departments from marketing to customer service to make decisions faster and bring competitive advantages to enterprises. Real-time data will become an indispensable tool for companies that want to make immediate use of insights, transforming analysis from a temporary and retrospective tool to a driving force that actively promotes business development.

-- Starburst co-founder and CEO Justin Borgman

accelerate and expand AI through data products

clear Data products will be the prerequisite for expanding AI workflows (such as RAG). As we all know, the effect of AI depends on the quality of input data, and the importance of data quality and governance will be more prominent. In addition, data products contain business contexts, which are critical for AI applications.

-- Starburst co-founder and CEO Justin Borgman

automated Data Pipeline

in 2025, a real automated data pipeline will be launched to eliminate manual intervention in data workflows. This breakthrough will enable the technical team to focus on innovation while ensuring consistency and high quality of large-scale data delivery. Real-time processing capabilities will become standard, making responsive applications that were difficult to implement in the past possible.

-- Anil Inamdar, head of consulting services, NetApp Instaclustr

the unstructured data governance process in the AI field will become increasingly mature.

Protecting enterprise data from leakage and misuse, while preventing AI from producing bad or wrong results, is the focus of today's executives. The lack of uniform standards, guidelines and regulations in North America makes this task even more difficult. IT leaders can start with data management technology. The first task is to obtain the visibility of all unstructured data in their storage. This visibility is the starting point for a better understanding of the increasing amount of data so that it can be properly managed and managed. Data classification is another key step in AI data governance. It involves identifying sensitive data through metadata markers, which cannot be used in AI programs. The enrichment of metadata can also help researchers and data scientists search for file content through keywords and quickly sort out the required data sets. Through the automated data classification process, IT can create workflows to continuously send protected datasets to secure locations and AI-ready datasets to object storage for AI tools. Automated data workflow orchestration tools will play an important role in the efficient management of PB-level data assets. The AI-ready unstructured data management solution will also provide a way to monitor workflow progress and audit risk results.

-- Krishna Subramanian, co-founder and chief operating officer of Komprise

the evolution of the storage administrator role: embracing security and A I data governance

under the dual pressure of data security and AI, the role of storage IT professionals is changing significantly. Today, storage management technology has become more automated and self-healing, and cloud-based systems are easier to manage. At the same time, the interactivity and interdependence among network security, data privacy, storage and AI are increasing day by day. Storage professionals need to ensure that data can be easily accessed and classified for AI use, and work with other functional departments to develop data governance plans, to deal with ransomware and prevent misuse of enterprise data in AI. The storage team also needs to master the location of sensitive data and have the tools to develop audited data workflows to prevent sensitive data leakage.

-- Krishna Subramanian, co-founder and chief operating officer of Komprise

centralized data: a powerful tool to optimize decision-making and break down departmental barriers

centralized information access will become the core element of seamless data flow, thus supporting real-time decision-making and interdisciplinary collaboration. As product design evolves from basic mechanical structure to complex integrated system (integrating electronics and software), collaboration among engineering, design and manufacturing teams becomes more and more critical. A unified view of real-time data can eliminate isolated information islands, accelerate decision-making processes, and stimulate innovative vitality in product development. Realizing the priority of data integration access will effectively alleviate the limitations brought by the "Partition mentality", which leads to the team's own affairs and the information transmission between departments is often one-sided. In the future, a unified and real product data source will emerge as the times require, enabling the engineering and manufacturing teams to collaborate from the early stage of design, optimize the communication process and accelerate decision-making. Research shows that nearly half of manufacturers are still suffering from data dispersion and isolation, resulting in impaired decision quality. By 2025, this proportion urgently needs to drop significantly. With integration tools, real-time centralized data access, and interdisciplinary collaboration, enterprises will shorten the time to market, enhance innovation capabilities, and avoid high error costs caused by information dispersion, so as to lead the future development of the design and manufacturing field.

-- Manish Kumar, CEO and Vice President of R & D of SOLIDWORKS

independent AI agents and the rise of large quantitative models

enterprises will tend to adopt the agent strategy with AI as the core to solve the problem, that is, to build an AI system that can make independent decisions based on environmental interaction. However, these agents depend not only on language Models, but also on Large Quantitative Models (LQMs). LQMs will integrate massive amounts of quantitative data and combine the physical perception architecture to meet the needs of diversified use cases. It is expected that there will be disruptive changes in the fields of drug research and development, material design, medical diagnosis, financial modeling and industrial optimization.

-- Dr. Stefan Leichenauer, vice president of SandboxAQ engineering

generative AI: unlock unstructured data value and accelerate intelligent business decision-making

the core of generative AI is to unlock the value of unstructured data, which constitutes the main body of enterprise information. By converting these data into actionable insights, AI can significantly improve enterprise decision-making efficiency and reduce dependence on manpower in complex analysis.

-- Michael Curry, president of Rocket Software data modernization

Data Quality: The cornerstone of AI-enabled service roles

the rapid iteration and application of AI tools are reshaping the service-based role, and their direct impact is reflected in customer support, IT support, and marketing functions. As the organization moves from the experimental stage to a more mature process, these service roles will evolve into a more robust and available process in 2025. IT organizations will deploy and standardize AI empowerment services more widely, and IT support roles will extend to the deployment and optimization of AI tools beyond the help desk functions. High quality data is the key to the successful transformation of these service roles. Good data organization and clarity are essential to ensure that AI tools effectively assist customer queries and service delivery.

-- Julie Irish, senior vice president and chief information officer of Couchbase

generative AI: transforming data Cemetery into AI gold mine

many organizations are deeply trapped in "data cemeteries", and these warehouses that store historical information are idle due to high maintenance or analysis costs. High data marking and tracking costs are the main causes of this phenomenon. Many enterprises adopt the strategy of "storing everything and analyzing little" due to the complexity and high cost of data management. However, a large number of valuable insights are still hidden in emails, documents, customer interactions, and operational data many years ago. The generative AI tool provides an unprecedented opportunity to efficiently process and analyze unstructured data. Enterprises can explore historical trends, customer behavior and business models, which were once incapable of analysis due to complexity. Unstructured data that was difficult to use in the past will become a valuable resource for training AI models in specific fields.

-- Haseeb Budhani, co-founder and CEO of Rafay Systems

analysis and Operation: The road of integration in 2025

by 2025, the artificial boundary between analysis and operation will disappear, because enterprises realize that the real commercial value comes from the integration of the two. This integration will be reflected in data products that optimize real-time operations and directly improve customer experience, such as complex recommend engines that can change according to business conditions.

-- Anil Inamdar,NetApp Instaclustr Consulting Service director

machine learning and data synthesis: a new impetus to Accelerate Supply Chain Innovation

looking forward to 2025, machine learning will be revitalized in the supply chain field. Although we have been able to interpret visual data from cameras, sensors and IoT devices, the addition of synthetic data will enable us to create diversified high-quality training sets and optimize the model before it goes online. For example, imagine an automated forklift in a warehouse to avoid potential dangers. By synthesizing data, we can simulate the scene when a vehicle hits a storage rack or crowd, making the model more robust before the production stage and avoiding damage and damage without actually experiencing these situations.

-- James Brenan, director of global consulting, Endava

revenue data: the ultimate asset of an enterprise

by 2025, income data will become the core asset of enterprises. The ability to integrate, analyze and utilize internal and external data streams will determine the position of market leaders. In an era of insight-driven actions, enterprises that fail to fully utilize their data potential-whether through advanced analysis or AI-will face elimination. Using income data to promote market strategy and implementation will become a magic weapon for the next generation of subverters.

-- Andy Byrne, CEO of Clari

data protection policy: the evolution of data security

by 2025, the data protection policy will shift from focusing on the security of static data or transmitted data to protecting data in use. Privacy protection technologies (similar state encryption and confidential Computing) will be widely used due to increased compliance requirements and real-time collaboration requirements. Industries such as medical care and education will protect their rich personal and organizational data resources through AI-driven anomaly detection technology to cope with the increasing attention of attackers to these industries. Event Response will shift from annual desktop drills to continuous testing through simulated attack platforms, enabling organizations to evaluate their response capabilities in real time.

-- Adam Khan, vice president, Global Security Operations, Barracuda

AI: a leap from insight to strategic business value

looking forward to 2025, AI will change from providing insight to realizing strategic business value through contextualization. With the further integration of data and analysis in the coming year, workflow will be able to connect the organization's real-world event, operation and personnel data to understand the specific roles of employees and teams. This means that AI will go beyond a wide range of data output to generate actionable insights. Employees at all levels will be empowered through personalized information to make better decisions and provide better results for customers. The leap in productivity achieved by enterprises through situational AI will be obvious. Customization based on real-time data and insight will become the key to success in an increasingly competitive environment.

-- John Licata,ServiceNow innovation officer

microscopic perspective of data labeling sources

in the technical field, discussions on how to obtain appropriate data sets and their labeling methods have never stopped. In fact, these labeling work is usually carried out in the form of outsourcing on a global scale, especially in developing countries, whose working conditions and remuneration level are often questioned. For example, task-based workers may need to process hundreds of thousands of images and get paid according to the number of accurate classifications. At the same time, although the demand for AI engineers is extremely high and the salary far exceeds the average market level, the current situation of this "sub-economy" still causes many problems.

-- Gordon Van Huizen, senior vice president of Mendix strategy

digital transformation of medical industry will improve data quality

in the coming year, digital transformation will continue to advance in the medical industry, ultimately improving the overall data quality and broadening the application scenarios of AI in the industry. At present, medical institutions are using AI to simplify work processes, reduce the administrative burden of employees, and optimize manual tasks such as scheduling. As organizations continue to introduce new digital tools and upgrade existing technology stacks, they will have more capabilities to manage, analyze, and clean up valuable data in a wider range.

-- Dr. Hugh Cassidy, head of LeanTaaSAI and chief data scientist

the ability to interpret, analyze and use unstructured data will determine the success or failure of an enterprise.

As the name implies, unstructured data is often the most difficult to interpret, but it is one of the most valuable information in enterprises. According to Gartner, 80% of enterprise data is unstructured. In 2025, the distinction between "owners" and "non-owners" will become more and more obvious: those enterprises that have mastered the tools and technologies for processing unstructured data and adapting it to AI, it will take a leading position in the new intelligent era. These companies can not only generate deeper business insights, but also their teams and agents have better decision-making ability, from analyzing customer emotions to writing blog articles to formulating competition strategies.

-- Sarah Walker, chief operating officer of Slack

from "arms race" to "data race"

over the past year, the AI boom has attracted a large number of investors and enterprises. However, the real winners are those enterprises that surpass "hot words" and focus on actual results and organizational needs. Nowadays, the core of the AI boom is not the technology itself, but the data value provided by AI. In 2025, enterprises that adopt pragmatic AI strategies and their underlying data infrastructure will perform best in promoting new insights and discoveries. Leading enterprises in the "data and algorithm competition" will not only use each piece of collected data to achieve differentiated AI results, but also through efficient management, organization, indexing and classification processes greatly surpass competitors.

-- Skip Levens, product director and AI strategist of Quantum Media and Entertainment Department

entering a new era of digitalization and data governance

as more enterprises deploy GenAI in 2025, they will enhance the return on investment and improve business results by strengthening data governance and digitization. Data governance is the cornerstone for the success of generative AI applications, ensuring that AI systems can run efficiently and provide useful and unbiased insights. At the same time, enterprises need to solve bottlenecks such as data cleaning and labeling. According to the Appian report, the obstacles caused by these problems increase by 10% every year. In addition, digitization will help enterprises standardize data and improve data consistency, thus promoting collaboration and decision-making between employees and stakeholders based on the same data set. Combining AI governance platforms and strategies, these advances will create more value for Enterprise's generative AI applications in 2025 and beyond.

-- Scott Francis,PFU America Technology preacher

automated development of data Observability

nowadays, the observability of data has reached a certain market maturity, and automation has become the key to maximize its value. In the future, observability tools will pay more attention to shortening users' operation time on the platform through automated deployment, problem identification, triage and resolution processes. With the gradual standardization of best practices, accelerating these processes will become the key to achieving the actual return on investment and help the team solve data problems with minimal manual intervention.

-- Egor Gryaznov, chief technology officer, Bigeye

AI will promote the importance of data quality in model Training and Analysis

AI refocuses on data quality mainly for two reasons: first, high-quality data is a prerequisite for training and fine-tuning models; second, AI-driven analysis tools can reveal undiscovered quality problems in data with higher resolution.

-- Ryan Janssen, CEO of Zenlytic

unstructured data management solutions to meet AI data governance and monitoring needs

according to Komprise's 2024 unstructured data management status report, IT leaders regard AI data governance and security as the primary capabilities of future solutions. AI data governance covers preventing data leakage or abuse, ensuring compliance, managing data deviations, and preventing AI from generating false or misleading results. Monitoring and alarming of capacity problems or exceptions are still important requirements, and the analysis and reporting functions have also attracted much attention. IT and storage owners will seek unstructured data management solutions that can automatically protect, segment, and audit sensitive and internal data usage, especially when AI is maturing, this use case will continue to expand.

-- Krishna Subramanian, co-founder and chief operating officer of Komprise

hybrid cloud persists, requiring deep data and cost insight

after cloud priority policies, cloud migration, and repeated operations, the survey shows that hybrid clouds will continue to exist in the foreseeable future. IT leaders recognize that hybrid use of local, edge, and cloud computing is a smart and low-risk strategy to meet the needs of different workloads and departments. Storage and cloud providers will adapt to this reality, and IT needs to have a deep understanding of its data assets in order to move IT to the optimal storage location in the data lifecycle. Optimizing a hybrid cloud storage environment is a dynamic goal that requires analysis based on real-time data types, growth, and access patterns, as well as flexibility in moving data to secondary or cloud storage layers as needed.

-- Krishna Subramanian, co-founder and chief operating officer of Komprise

global Unified namespaces do not win single

unstructured data is stored in multiple places, and these data Islands make it difficult to manage and mine their value. Many storage providers try to provide a unified solution and a single global namespace by allowing customers to migrate completely to their platforms and stay away from other isolated islands. However, this simplified assumption is not realistic. Customers use multiple storage providers and architectures because data needs vary in their lifecycle. In addition, experience over the past decade shows that customers tend to maintain a hybrid model and take advantage of a combination of local and cloud services. The key to solving the problem of isolated islands is not to eliminate isolated islands, but to cross all isolated islands through a unified view, supporting two-way data conversion between files and objects, and expanding the main namespace.

-- Krishna Subramanian, co-founder and chief operating officer of Komprise

data tools will better meet diverse business needs

when enterprises initially deployed data management tools, their main goal was to centralize and orderly the large amount of data they owned but difficult to manage. They want to track Data from the central Center, which gives birth to Data Lake and Data Warehouse. However, as many organizations solve this chaotic problem, they are now turning to data management tools to seek more complex and sophisticated uses. Enterprises want to provide the required data for each department and provide it in the required way. This requires a more complex and decentralized Data management method that relies on tools such as Data Mesh and Data Mart. Central data repositories will not disappear, but they will increasingly exist in parallel with data tools and platforms that can better meet diversified business needs.