Gartner: Top Strategic Technology Trends for 2025 PDF

Top Strategic Technology Trends for 2025 21 October 2024 - ID G00815761 - 40 min read By Analyst(s): Gene Alvarez, Tom Coshow, Jasleen Kaur Sindhu, Dan Ayoub, Mark Horvath, Nick Jones, Soyeb Barot, Frank Buytendijk, Marty Resnick, Bill Ray, Sylvain Fabre, Moutusi Sau, Bart Willemsen Initiatives:Technology Innovation and Strategy We’ve identiﬁed the 10 strategic technology trends that will have the most impact in the next ﬁve years and beyond — trends that span AI imperatives and risks, new frontiers of computing and human-machine synergy. Tracking these will help IT leaders shape the future with responsible innovation. Overview Opportunities Agentic AI brings imperatives and risks and will enable organizations to transform the nature and efﬁciency of work, processes and decision making. However, it will also drive the need for advancements in AI governance technology. The technology created to defend organizations from the effects of disinformation will protect people, organizations and society. New frontiers of computing keep expanding the potential for beneﬁt but also bring threats. Quantum computing will break today’s cryptography, exposing everyone to risk. Tiny, ultra-low-cost wireless tags and sensors will enable new business models and ecosystems. New energy-efﬁcient compute models will meet the demand for more computing and sustainably. The growing numbers of computing models provide an opportunity for integration and orchestration to optimize the use of all models. Human-machine synergy is increasing, with the creation of next-level interactions between physical and virtual experiences. Robots that perform more than one function will integrate into humans’ daily lives. Technology will bring the ability to directly access and improve thoughts and emotions for enhancing human cognition and performance, and bringing about new ways to help people thrive. Gartner, Inc. | G00815761 Page 1 of 30 This research note is restricted to the personal use of [email protected]. Recommendations Identify opportunities to add agentic AI to workﬂows where signiﬁcant demand for scale and efﬁciency exists and adaptability is required. Ensure fairer AI systems by considering multiple perspectives when designing and evaluating AI methods. Include deepfake detection in systems such as identity veriﬁcation. Develop policies to ease the transition to new cryptographic algorithms. Identify information blind spots and early opportunities to collect data from your physical environment using ambient intelligence. Use computing more efﬁciently by switching to greener cloud providers. Manage the complexity of using diverse computing models building a robust and scalable orchestration layer for provisioning and managing resources. Invest in the necessary infrastructure for spatial computing. Adopt a policy of “polyfunctionality by default” for all robot deployments. Set up proofs of concept for neurological enhancement solutions where you have applicable use cases. What You Need to Know Shape the Future With Responsible Innovation Organizations face the challenge of continually having to innovate to meet business challenges and disruptions. As new technologies arise, they present many opportunities but increasingly bring ethical challenges and considerations. Organizations must act responsibly to balance innovation while retaining the trust of their customers, employees and partners. This research will help you shape the future for your organization with responsible and ethical innovation. AI imperatives and risks abound as organizations move forward with AI agents. This, combined with other aspects of AI, will drive a need for AI governance platforms within organizations, enabling all to use AI responsibly and ethically. Malicious actors using AI to accelerate the spread of disinformation can cause signiﬁcant damage to an organization, its customers, partners and employees. Enterprises will need technologies to track the spread of information by, or about, their organization to assess the truth of that information and create trust. Organizations must also protect themselves from malicious actors using synthetic media to gain real-time access to their systems and spread misinformation. Gartner, Inc. | G00815761 Page 2 of 30 This research note is restricted to the personal use of [email protected]. New frontiers of computing are being created, requiring organizations to look differently at how they compute. As new security measures will be needed, information in the shadows today must become visible in the future. Organizations will need to meet growing compute demand while lowering their carbon footprint. They also must integrate and orchestrate many compute models, operating them as one in the most efﬁcient way to meet their rising computing needs. In these new frontiers of computing, quantum computing threatens to break today’s cryptology, exposing everyone to risk. A new cryptology is needed to protect organizations and society. Tiny, ultra-low-cost wireless tags and sensors will make possible real-time, large-scale tagging, tracking and sensing — enabling new business models and ecosystems. The increasing demand for computing and the lack of energy to support it drives the need for new energy-efﬁcient compute models. The optimization of the growing numbers of new computing models working with all existing models will push organizations to focus on integration and orchestration of computing. Advances in the way humans and machines work together are creating a new level of human-machine synergy. The creation of next-level interactions between physical and virtual experiences will bring together the physical and digital world through spatial computing. Humans will have robots working side-by-side with them in the same environment and even can become teammates. Robots’ ability to perform more than one function will integrate them into humans’ daily work and home experience. Humans will become integrated with machines through wearable or implanted technologies, and neurological enhancement will give us the ability to directly access and improve thoughts and emotions. This will enhance human cognition and performance, bringing about new ways to help humans. Wearables and implanted technologies, along with polyfunctional robots, will forever change how humans and machines work together, moving us into a world where all these technologies exist to beneﬁt humans. Gartner, Inc. | G00815761 Page 3 of 30 This research note is restricted to the personal use of [email protected]. Figure 1: Top Strategic Technology Trends for 2025 Trend Profiles. Click links to jump to profiles AI Imperatives and Risks New Frontiers of Human-Machine Synergy Computing Agentic AI Postquantum Cryptography Spatial Computing AI Governance Ambient Invisible Polyfunctional Robots Platforms Intelligence Disinformation Security Energy-Efficient Computing Neurological Enhancement Hybrid Computing AI Imperatives and Risks Agentic AI Back to top Analysis by Tom Coshow, Gary Olliffe Strategic Planning Assumptions: Gartner, Inc. | G00815761 Page 4 of 30 This research note is restricted to the personal use of [email protected]. By 2028, 33% of enterprise software applications will include agentic AI, up from less than 1% in 2024. By 2028, AI agent machine customers will replace 20% of the interactions at human- readable digital storefronts. By 2028, at least 15% of day-to-day work decisions will be made autonomously through agentic AI, up from zero percent in 2024. Agentic AI systems autonomously plan and take actions to meet user-deﬁned goals. Current AI assistants and large language models (LLMs) perform tasks including generating text, summarizing content or basic use of tools, but they haven’t been able to take action on their own “initiative.” Instead, they’ve acted on users’ prompts or followed orchestrated processes, but agentic AI is changing that. It offers the promise of a virtual workforce of agents that can assist, ofﬂoad and augment human work or traditional applications. The goal-driven planning capabilities of agentic AI also promises to deliver more adaptable software systems, capable of completing a wide variety of “undeﬁned” tasks within a domain, rather than only those designed into the software. AI agency is a spectrum. It runs from traditional systems with limited agency that perform speciﬁc tasks under narrowly deﬁned conditions to future agentic AI systems with full agency that learn from their environment, plan approaches, make decisions and perform tasks independently. Emerging and existing AI features, such as agent and multiagent frameworks, AI guardrails and function-calling, enable systems that use LLMs to have agency. Agentic systems using these capabilities can plan, act and adapt based on their context to meet goals in complex environments, dramatically increasing AI’s potential. For example, agentic AI can examine data, do research, compile a list of tasks to complete and then take those actions in the digital world or physical world via APIs or robotic systems. Agentic AI, with its ability to take action autonomously or semiautonomously, has the potential to realize CIOs’ desire to increase productivity across the organization. 1 This motivation is driving both enterprises and vendors to explore, innovate and establish the technology and practices needed to deliver this agency in a robust, secure and trustworthy way. Gartner, Inc. | G00815761 Page 5 of 30 This research note is restricted to the personal use of [email protected]. The key challenges facing organizations that are building early AI agents lie in the need to establish high levels of trust and conﬁdence. Agents that can autonomously select and use tools (via features APIs and function calling) must be constrained by robust guardrails that ensure their actions are aligned to the provider’s intentions and properly reﬂect user intent. Frameworks for enforcing AI guardrails are growing in maturity, but commonly also rely on LLMs as part of their processing, leaving a margin for error. Additionally, the behavior of AI agents that may formulate unique plans for each request, goal or scenario makes testing and validating their behavior much more challenging than both traditional software and more mainstream applications of generative AI (GenAI). Finally, monitoring and governing the behavior of AI agents in deployment requires new techniques and tools, since their behavior will change and adapt as they memorize both shared and user-speciﬁc context over time. The deployment of AI agents that adapt overtime must come with a security plan that monitors the AI agents and has forward- looking guardrails. Agentic AI is at the forefront of current R&D efforts by many major vendors in the AI marketplace. The technology is on the cusp of being capable of moving from the lab to early, lower risk, innovation-tolerant use cases (e.g., in accelerating software engineering beyond the capabilities of existing AI code assistants). However, the leap from the lab to production for more business-critical applications and processes that would enable operational efﬁciencies and improvements at scale is not a trivial one. It is entirely dependent on the emergence of reliable and predictable patterns, practices and technologies for delivery, supporting and governing a virtual workforce that can plan, act and adapt over time. Actions: Identify opportunities to add agentic AI to workﬂows where signiﬁcant demand for scale and efﬁciency exists and adaptability is required. Rethink entire workﬂows across silos from an automation-only perspective and add humans back into new workﬂows at strategic points. Start small on use cases where high-quality data is accessible and behavior is veriﬁable. Treat AI agents like Tier 1 digital co-workers that you delegate work to. Rethink collaboration models, workﬂows and team strategies to maximize the beneﬁt of AI agents that can uncover and act on derivative events that human teammates might not notice. Gartner, Inc. | G00815761 Page 6 of 30 This research note is restricted to the personal use of [email protected]. Put in place guardrails to ensure agentic AI is constrained to a deﬁned role and set of capabilities. Do so to prevent it from taking incorrect actions that cause damage. For more information, see Top Strategic Technology Trends for 2025: Agentic AI. AI Governance Platforms Back to top Analysis by Jasleen Kaur Sindhu, Moutusi Sau, Svetlana Sicular Strategic Planning Assumptions: By 2028, enterprises using AI governance platforms will achieve 30% higher customer trust ratings and 25% better regulatory compliance scores than their competitors. By 2028, organizations that implement comprehensive AI governance platforms will experience 40% fewer AI-related ethical incidents compared to those without such systems. AI governance platforms are technology solutions that enable organizations to manage the legal, ethical and operational performance of their AI systems. These platforms, part of Gartner’s evolving AI Trust, Risk and Security Management (AI TRiSM) framework, help enforce enterprise AI policies. They offer key features such as creating, managing and enforcing policies for responsible AI use, explaining how AI systems work, model life cycle management and providing transparency to build trust and accountability. AI governance platforms also assess risks related to data quality and privacy violations, offering continuous model monitoring and correction to ensure fairness. Additionally, these platforms guide AI models through structured processes, track usage, monitor performance and help organizations comply with regulations like the General Data Protection Regulation (GDPR), California Consumer Privacy Act (CCPA) and AI regulations. AI’s rapid integration into various sectors — especially highly regulated industries — has highlighted the urgent need for AI governance platforms. As AI technologies become more widespread and sophisticated, the risks of bias, lack of transparency and data privacy concerns grow. Increasing global regulations on AI use and data privacy push organizations to adopt robust governance methods. Gartner, Inc. | G00815761 Page 7 of 30 This research note is restricted to the personal use of [email protected]. For example, the EU’s AI Act 2 requires high-risk AI systems to respect human rights, accelerating the demand for comprehensive AI governance platforms. Public concern about AI harms is also growing, with a 2023 Pew Research Center 3 survey ﬁnding that 52% of respondents are more concerned than excited about AI. High-proﬁle cases of AI bias and disinformation have further highlighted the need for better oversight and governance. Despite the beneﬁts, implementing AI governance platforms comes with challenges. AI governance practices are still emerging, with varying guidelines across regions and industries. The fast pace of AI development makes it hard to establish consistent and future-proof governance practices. Cultural barriers and resistance to change can also impede adoption. However, there are signiﬁcant opportunities. Technological advancements in explainable AI, automated model monitoring and scalable AI governance tools can help organizations implement robust governance practices. Adopting AI governance platforms can help organizations stay compliant with evolving regulations, reduce legal risks and build trust. Transparent and fair AI systems can improve consumer trust and brand reputation, providing a competitive edge. There are several emerging examples of ﬁrms who have adopted AI governance practices as a part of their AI operations. For example, HSBC has implemented its Principles for the Ethical Use of Data and AI by establishing governance frameworks that prioritize transparency and accountability, regularly checking for fairness and bias, protecting customer privacy and training employees on ethical practices. 4 Similarly, Unilever has an AI governance framework that includes an AI assurance compliance process, involving internal reviews, third-party assessments and continuous monitoring to ensure AI systems align with ethical standards and operational goals. 5 Actions: Ensure multiple perspectives are considered when designing and evaluating ethical and responsible AI methods, leading to more balanced and fair AI systems. Involve a diverse range of stakeholders, including ethicists, technologists and affected communities in the AI governance process. Deﬁne clear lines of responsibility and accountability for AI-related decisions and actions. This will enhance transparency and trust, making it easier to address issues when they arise. Gartner, Inc. | G00815761 Page 8 of 30 This research note is restricted to the personal use of [email protected]. Perform regular audits of AI systems to identify and mitigate risks to ensure continuous alignment with governing standards and regulatory requirements. Allocate resources to develop in-house responsible and ethical AI expertise. Dedicated AI governance teams are more likely to scale AI initiatives successfully and responsibly. For more information, see Top Strategic Technology Trends for 2025: AI Governance Platforms. Disinformation Security Back to top Analysis by Dan Ayoub, Akif Khan Strategic Planning Assumption: By 2028, 50% of enterprises will begin adopting products, services or features designed specifically to address disinformation security use cases, up from less than 5% today. Disinformation security is an emerging category of technologies aimed at systematically discerning trust. Its objective is to provide methodological systems for ensuring integrity, assessing authenticity, preventing impersonation and tracking the spread of harmful information. Primary use cases for enterprises include: Detecting use of synthetic media in unauthorized contexts (i.e., identity veriﬁcation, real-time communications or claims validation). Intelligence monitoring for narratives spread through mass or social media (i.e., those targeting an executive leadership team, products, service or brand). Preventing the impersonation of individuals doing business with an organization (i.e., employees, contractors, suppliers and customers). Disinformation security expands traditional enterprise defenses through use of emerging technologies by: Gartner, Inc. | G00815761 Page 9 of 30 This research note is restricted to the personal use of [email protected]. Strengthening controls for validating identity and ensuring integrity of real-time communications within workforce and consumer applications to prevent fraud. Deepfake detection can be integrated into workﬂows for identity, authentication, claims validations, messaging applications and phone systems. Mitigating account takeover through continuous risk scoring with contextual awareness across session and event data within an online user journey. Impersonation prevention extends beyond authentication by leveraging a continuous adaptive trust model for risk assessment. Protecting brand reputations by proactively identifying harmful narratives, how they are spreading and their source for strategic incident response communications. Reputation protection includes the ability to proactively identify harmful narratives targeting a brand, product or executives. Disinformation is already a digital arms race: phishing, hacktivism, fake news and social engineering are all being turbocharged by adversaries intent on sowing fear, spreading havoc and committing fraud. Given the wide availability and advanced state of artiﬁcial intelligence and machine learning tools being leveraged for nefarious purposes, the number of disinformation incidents targeting enterprises is expected to increase over time. Left unchecked, disinformation can cause signiﬁcant and lasting damage to any organization. Disinformation security has come to the fore because of the following recent developments: Misinformation and disinformation were recognized as the top global threats for the next two years by the World Economic Forum in 2024. As more organizations are targeted with disinformation attacks, hype and concern around this category will continue to grow. Threat actors attempt to manipulate public opinion at critical times of importance through social media inﬂuence operations and fake news websites. Elections are being conducted around the globe throughout 2024, with disinformation security playing a central theme. Deepfake attacks are increasing rapidly, with as many as one in 10 executives already reporting their businesses being targeted. Biometric authentication was thought to be safe, accurate and reliable, but deepfakes show that no single technology layer is impenetrable. Gartner, Inc. | G00815761 Page 10 of 30 This research note is restricted to the personal use of [email protected]. Harmful narratives are being ampliﬁed with AI to quickly spread online, damaging reputations and manipulating audiences before organizations are prepared to respond. Enterprises are now adopting narrative intelligence solutions pioneered for government and defense use cases. Actions: Ensure deepfake detection is covered as a feature capability within broader systems such as identity veriﬁcation, biometric authentication and claims validation. Look for dedicated solutions to authenticate third-party content or monitor integrity of real- time workforce communications systems. Move beyond authentication to incorporate impersonation prevention through continuous adaptive trust models that continuously assess risk across the entire user journey. Look for context-aware solutions that incorporate behavioral analytics within probabilistic orchestration models. Incorporate digital risk protection services, narrative intelligence and media monitoring as part of a layered brand safeguarding strategy. Leverage tools to proactively identify information harmful to the organization and cross-functional teams to develop response strategies. For more information, see Top Strategic Technology Trends for 2025: Disinformation Security. New Frontiers of Computing Postquantum Cryptography Back to top Analysis by Mark Horvath, Bart Willemsen Strategic Planning Assumption: By 2029, advances in quantum computing will make most conventional asymmetric cryptography unsafe to use. Gartner, Inc. | G00815761 Page 11 of 30 This research note is restricted to the personal use of [email protected]. In 2019’s Top Strategic Technology Trends, we pointed at a long-term trend to prepare for: quantum computing (QC). Ever since, QC developments have progressed steadily and expectations of its availability before the end of 2020 proved on point. QC is expected to mean the end of several types of conventional cryptography used widely in billions of devices and over 80% of communications over the global internet. Criminals and state actors show an understanding of this importance as they have recently demonstrated attack scenarios such as “harvest now, decrypt later.” In these attacks, encrypted data is stored after exﬁltration, with the expectation of being able to decrypt these stolen secrets, data and other sensitive information in the future when decryption by a quantum computer becomes more approachable. Postquantum cryptography (PQC) provides data protection that is resistant to QC decryption risks. However, switching cryptography methods in existing architectures is not an easy task, as the new algorithms are not drop-in replacements for existing asymmetric algorithms. As such, organizations must prepare a longer lead time to ready themselves for robust protection of anything sensitive or conﬁdential. Owners of sensitive assets will not know if their data has been harvested and later decrypted until it’s too late. These are not just risks to the business but to any individual or group of individuals whom that data may be about as well. The approach toward PQC involves creating a repeatable process collectively referred to as “crypto-agility.” Crypto-agility is the capability to transparently swap out encryption algorithms and related artifacts in an application, replacing them with newer, presumably safer, algorithms. A consistent approach is needed, because, given past and future developments in cryptography, this will not be the last time we have to switch encryption methods. Adopting PQC won’t be easy. No drop-in alternatives exist for current cryptographic algorithms. This means discovery, categorization and reimplementation efforts will be necessary. Additionally, new algorithms have different performance characteristics from non-PQC ones. For example, key and ciphertext sizes are larger, and encryption and decryption times are longer, which may impact performance. This means current applications must be retested and, in some cases, rewritten. Actions: Develop policies to ease the transition to new algorithms. Do so because adopting a policy-based program will reduce confusion and arbitrary choices, and increase manageability. Gartner, Inc. | G00815761 Page 12 of 30 This research note is restricted to the personal use of [email protected]. Build a cryptographic metadata database of all in-use cryptographic algorithms. Use it to identify the expected end-of-life targets in the short-, mid- and long-term time scales. Create a key life cycle policy to reﬂect the risks to asymmetric keys. Implement crypto-agile application development and move to production after extensive testing. Vet and test new PQC algorithms to understand their characteristics, uses and performance. Upgrade or replace hardware where necessary. For more information, see Top Strategic Technology Trends for 2025: Postquantum Cryptography. Ambient Invisible Intelligence Back to top Analysis by Nick Jones Ambient invisible intelligence is enabled by ultra-low-cost, small, smart tags and sensors, which will deliver large-scale affordable tracking and sensing and, in the long term, will enable a deeper integration of sensing and intelligence into everyday life. Ambient invisible intelligence is driven by three key technologies: Low-power wireless, low-energy Bluetooth will dominate in 2025, but technologies such as Wi-Fi and 5G are also exploring extensions to support ambient wireless. Innovations, such as backscatter wireless, will likely also be used in the future. Energy harvesting, especially from ambient RF energy. This enables small battery- free tags and sensors with effectively inﬁnite life span and eliminates the size and cost of battery components. Low-cost, low-energy electronics, enabling chips that are efﬁcient enough to use harvested power to run sensors, perform simple computation and send short-range wireless messages. Gartner, Inc. | G00815761 Page 13 of 30 This research note is restricted to the personal use of [email protected]. Through 2028, early examples of ambient invisible intelligence will focus on solving immediate problems by enabling low-cost, real-time tracking and sensing of items to improve visibility and efﬁciency. Such examples will be limited to single organizations or tightly coupled supply chains because of current systems’ need for infrastructure. Examples of processes that could beneﬁt include retail stock checking, perishable goods logistics and organizations managing large numbers of packages, such as couriers and postal services. In the longer term, we expect that ultra-low-cost electronics will remain in items throughout their life. The messaging formats involved will be standardized, and the gateways to receive ambient messages will become an integrated feature of homes and ofﬁces. For example, this could occur as part of Wi-Fi access points or while embedded into consumer white goods. This will enable new ecosystems, like smart packaging communicating with domestic refrigerators and clothing items communicating with washers and dryers. Small intelligent tags associated with objects could also provide unforgeable provenance and new ways for objects to report on their identity, history and properties (e.g., to satisfy EU requirements for digital product passports). In the short term, ambient devices will focus on communications, location and simple environmental sensing. However, as electronics improves and ambient concepts are incorporated into wireless standards, we expect device capabilities to grow to include simple intelligence and, perhaps, peer-to-peer communications. A side effect of ambient intelligence will be that much more real-time data will be available about processes, products and logistics to be used by analytics and AI for optimizing processes. Ambient invisible intelligence will also accelerate the trend to invisible analytics identiﬁed by Gartner in other research. However, privacy will be a major concern when small inconspicuous sensors and tags are embedded in objects for their entire life. Vendors must address privacy concerns and obtain consent for some types of data use. Some concerned consumers will likely want ways to physically disable or destroy tags and sensors. Actions: Identify information shadows and early use cases where ambient intelligence delivers return on investment in 2025 through 2028 and pilot the most promising examples. Consider this as a replacement for some current or planned RFID use cases. Gartner, Inc. | G00815761 Page 14 of 30 This research note is restricted to the personal use of [email protected]. Look for new AI and analytics opportunities enabled by real-time information (e.g., related to how objects are used, and how and where they are stored). Analyze the privacy implications of any proposed use of this technology and ensure that users can disable it if required (e.g., by physical destruction of the tag or veriﬁable deletion of keys). Collaborate with partners and industry-standards organizations to look for longer- term ecosystem opportunities based on lifelong knowledge of an object’s identity and properties. For more information, see Top Strategic Technology Trends for 2025: Ambient Invisible Intelligence. Energy-Efficient Computing Back to top Analysis by Nick Jones Organizations face growing legal, commercial and social pressure to improve their sustainability, and IT organizations must play their part. IT impacts sustainability in many ways. For example, water consumption by data centers, e-waste and recyclability are all important, but in 2024, the leading consideration for most IT organizations is their carbon footprint. Carbon is generated by developing and running applications, storing data and networking. Compute-intensive applications are likely to be the biggest contributors, which consume the most energy. Examples include AI training, simulation, optimization and media rendering. There are three levers that IT can use to control the carbon footprint of its systems: Architecture, code and algorithms. Develop applications with more efﬁcient algorithms, architectures and data structures. Hardware efficiency. More modern processors and disks are more efﬁcient. Special- purpose devices, such as graphics processing units (GPUs), are more efﬁcient than general purpose equivalents for some applications. However, replacing hardware incurs an embodied carbon cost, which may outweigh the efﬁciency improvements. Gartner, Inc. | G00815761 Page 15 of 30 This research note is restricted to the personal use of [email protected]. Greener power. Power runs IT systems and cools data centers. Carbon intensity is the amount of carbon emitted for each kilowatt hour of electricity generated. It will be almost zero for renewable sources and very high for fossil-fuel sources. Some IT organizations will choose when and where to run jobs to exploit regional differences in the supply carbon intensity. Short-term tactics include using greener energy, retiring inefﬁcient hardware, improving utilization or shifting jobs to greener cloud regions. In the medium term, good practices are emerging for efﬁcient coding, architecture and algorithms. Some algorithms may be ported to more efﬁcient hardware such as GPUs. Demand for IT, however, is inexorably rising — especially in areas with the largest carbon footprints, such as AI and optimization. In the long term, improvements of one to two orders of magnitude will be required, which is more than current technology can deliver. More radical approaches will be required, but in 2024, many are still academic research topics or, at best, early prototypes. Examples include neuromorphic systems, which can execute some types of AI and certain other tasks very efﬁciently. Starting in the late 2020s, we expect to see a number of optical computing systems emerge for special-purpose tasks, such as AI and optimization. These will perform some operations using dramatically less energy than their silicon counterparts (e.g., offering 100x or better improvements). In the long term, technologies such as DNA storage, ceramic storage and quantum computing also promise signiﬁcant sustainability beneﬁts. However, the path to more sustainable IT will be disruptive, because: New hardware, cloud services, skills, tools, algorithms and applications will be required. The enterprise hardware landscape will become more complex and diverse involving multiple new computing architectures and paradigms. Migrating to new computing platforms and architectures may be complex and expensive. Energy prices will likely rise in the short to medium term, especially for green energy where demand will outstrip supply in some regions. Gartner, Inc. | G00815761 Page 16 of 30 This research note is restricted to the personal use of [email protected]. Some organizations will face a rearchitecting sustainability chasm, where improved efﬁciency is required, but new technologies aren’t sufﬁciently mature. If IT leaders fail to improve sustainability, the performance of their business could be impacted. Sustainability constraints could limit their ability to deploy advanced solutions on current platforms. Actions: Adopt short-term tactical solutions, such as improving utilization, switching to greener cloud providers, shift workloads to greener cloud regions, run systems at times and locations where the local supply has a lower-carbon intensity. Identify current and planned systems that will have large carbon footprints. Consider rearchitecting and migrating some or all onto signiﬁcantly more efﬁcient hardware, such as GPUs. Monitor emerging technologies such as optical and neuromorphic computing and pilots when available. For more information, see Top Strategic Technology Trends for 2025: Energy-Efﬁcient Computing. Hybrid Computing Back to top Analysis by Soyeb Barot, Frank Buytendijk It’s often said that the future of computing is “quantum.” It’s not. The future of computing is hybrid. Hybrid computing combines different compute, storage and network mechanisms to solve computational problems. For example, it can combine neuromorphic, quantum, photonic and, eventually, bio and carbon computing technologies. It involves creating a hybrid environment built on an orchestration framework that uses the respective strengths and capabilities of the various mechanisms. Hybrid computing helps organizations explore and solve problems by combining specialized compute, storage and network mechanisms — combined. This helps technologies, such as AI, perform beyond current technological limits. Gartner, Inc. | G00815761 Page 17 of 30 This research note is restricted to the personal use of [email protected]. Hybrid computing will be used to create highly efﬁcient, high-speed transformative innovation environments. These environments will perform more effectively than conventional environments, leveraging the strengths and capabilities of each individual computation mechanism. They’ll be able to deal with high-dimensional optimization problems scalable for large, complex problems — using signiﬁcantly less energy. New use cases across multiple industries (i.e., manufacturing and logistics, ﬁnancial services, life sciences, materials and drug discovery) will focus mainly on two areas. First, higher levels of automation, working toward running complete autonomous businesses. Second, augmenting human capability, offering real-time personalization at scale and ultimately using the human body as a computing platform itself. However, hybrid computing requires exploring highly experimental technologies and scaling successful initiatives. Many of the involved technologies are not only nascent but highly complex, requiring specialized skills. Organizations seeking to create value from using hybrid computing must have a high tolerance for cost and challenges with complexity, security and trust. Organizations adopting hybrid cloud will have to rearchitect systems and applications to better integrate and interface across compute mechanisms. Application and system architectures will become more complex as they leverage more diverse hybrid compute mechanisms, amplifying the need for robust modularity and interface design. Hybrid compute environments will require a robust software, storage and network orchestration layer alongside associated services. They’ll need these elements to enable organizations to use multiple compute mechanisms for given portions of the application scope. Introducing hybrid computing poses signiﬁcant security risks, as each module part of the system will operate autonomously on computation and decision making. This means each module will generate data elements and pass them to other modules as part of the application workﬂow. The governance and security of data pipelines and overall systems will thus require an overhaul. Actions: Gartner, Inc. | G00815761 Page 18 of 30 This research note is restricted to the personal use of [email protected]. Manage the complex nature of a hybrid computing architecture that may easily become “messy” by building a robust and scalable orchestration layer. This may involve establishing delivery platforms that support distributed data management fabric, multiple software architectures and deployment of applications to diverse hybrid computing environments. Make sure that hybrid computing works by ensuring that distributed data management is in order. First, rethink data retention — decide what information is to be stored, where, how and for how long. Second, create a universal data fabric to generate and maintain all the metadata between the different data domains. Third, implement DataOps to manage data observability throughout the various data pipelines. Evaluate each computing mechanism for relevance in your environment and establish reference architectures to guide the adoption and integration of each — spanning data, infrastructure and application environments. Counter the computing cost increase over the coming years with hybrid computing. Do so to gain the opportunity to scale and achieve beneﬁts that far outweigh the cost of optimizing existing infrastructures. Create opportunity maps and rapidly scale performance by building efﬁcient and resilient autonomous computing environments. Do so because innovating while addressing climate sustainability will be vital as energy costs increase and compute energy consumption grows, largely driven by GenAI. For more information, see Top Strategic Technology Trends for 2025: Hybrid Computing. Human-Machine Synergy Spatial Computing Back to top Analysis by Marty Resnick Strategic Planning Assumptions: By 2026, 30% of manufacturing processes will use spatial computing to streamline and improve efficiencies by being more dimensionally accurate. Gartner, Inc. | G00815761 Page 19 of 30 This research note is restricted to the personal use of [email protected]. By 2027, over 40% of large organizations worldwide will use a combination of Web3, spatial computing and digital twins in metaverse-based projects aimed at increasing revenue. By 2028, 20% of people will have an immersive experience with persistently anchored, geoposed content once a week, up from less than 1% in 2023. By 2033, spatial computing will grow to $1.7 trillion, up from $110 billion in 2023. Spatial computing combines physical and digital objects in a shared frame of reference beyond screen-based displays. This involves spatial mapping (see Glossary key at the end of this document) and identiﬁcation of people, places and things within the physical world as a foundation for anchoring digital content that intersects with the physical world’s spatially anchored, indexed and organized content. Spatial computing digitally enhances the physical world with technologies such as augmented reality and virtual reality. It does so seamlessly, creating the next level of interaction between physical and virtual experiences. This expands the potential of physical and digital objects, and their monetization possibilities. Consumer demand is growing for immersive and interactive experiences in sectors such as gaming, manufacturing, education, ﬁnancial services and e-commerce. Additionally, the need for sophisticated visualization tools for better decision making and increased efﬁciency drives the uptake of spatial technology in industries such as healthcare, retail and manufacturing. Hype relating to spatial computing technologies and applications is increasing. This has been ampliﬁed by the introduction of new head-mounted displays (HMDs) that enable immersive digital experiences in the physical world. These devices include XREAL’s Air 2 and Beam Pro, 6 Apple Vision Pro 7 and Meta Quest 3. 8 Such devices have created the potential for new business models and opportunities for monetizing physical-digital interactions. In the next ﬁve to seven years, organizations’ use of spatial computing will increase their operational effectiveness through streamlined workﬂows and enhanced collaboration. It’ll offer real-time contextual information to users for improved decision making, especially in logistics and manufacturing. For example, spatial computing will enable: Gartner, Inc. | G00815761 Page 20 of 30 This research note is restricted to the personal use of [email protected]. New monetization models, such as the monetization of physical assets (e.g., users can click to buy event tickets, lease parking spaces or rent vehicles) Virtual collaboration for R&D, product design and training The overcoming of some accessibility issues — especially for those with visual impairment — through adaptive interfaces that can adjust text size, color contrast or enable screen readers to access content more comfortably Immersive experiences across product, sales and market funnels Interaction with digital twins in a manufacturing environment as part of industrial metaverse initiatives 9 Visual search for real-time information overlaid on physical products New consumer experiences that’ll increase market opportunities in consumer electronics, gaming and brand marketing More inclusive hiring because devices will enable employees to collaborate in a shared space, irrespective of their physical location However, several challenges exist to the uptake of spatial computing: Cost: HMDs are expensive, as is digitizing assets. Although phones and other mobile devices offer spatial computing experiences, functionality may be limited. Issues with HMDs: HMDs are heavy, making them uncomfortable. They also consume much battery power and require frequent charging. HMDs tend to isolate users from those around them. Concerns also exist about the potential for accidents when wearing HMDs. Complicated user interfaces: Many HMDs have more of an “onboarding process” with tutorials needed to fully understand actions that may be less than intuitive. Lack of a killer app in the consumer market: There is no application or use case driving long-term adoption. Today, there is more short-term interest with little stickiness. Device fragmentation: This remains an issue for developing and deploying spatial computing content. Gartner, Inc. | G00815761 Page 21 of 30 This research note is restricted to the personal use of [email protected]. Data privacy and security: Spatial computing uses cameras and sensors to collect data about the user’s environment, actions and interactions. Devices can store vast amounts of this personal data, and organizations must ensure this data’s safety. The lack of device management capabilities could also result in security vulnerabilities, but the sensitivity of stored data also poses ethical and legal issues. Actions: Invest in the necessary infrastructure for spatial computing, such as internet high- speeds, wireless, ubiquitous coverage, low latency and reliability, along with compatible devices. Assess where spatial computing could expand the utility and reach of your organization’s products and services. Identify speciﬁc use cases where spatial computing can add value for your organization. Evaluate, for example, the potential to enhance customer experiences, improve operational efﬁciency or facilitate remote collaboration. Adopt a platform to bring multiple spatial computing experiences together. Ensure it can optimize the streaming of content in areas with low bandwidth. For more information, see Top Strategic Technology Trends for 2025: Spatial Computing. Polyfunctional Robots Back to top Analysis by Bill Ray, Nick Jones Strategic Planning Assumption: By 2030, 80% of humans will engage with smart robots on a daily basis, up from less than 10% today. Task-speciﬁc robots, custom designed to repetitively perform a single function, are being replaced with polyfunctional machines capable of doing just about anything. These next- generation robots can take on a multitude of tasks, and seamlessly switch between them as required, improving efﬁciency and providing a faster return on investment — thus accelerating adoption across industries and use cases. Gartner, Inc. | G00815761 Page 22 of 30 This research note is restricted to the personal use of [email protected]. Not only will this accelerate adoption in manufacturing and warehouses, it also creates new opportunities for single-function robots that won’t make economic sense. Most consumer homes, for example, lack the space for multiple robots accomplishing different tasks, and it wouldn’t be economical to have such a ﬂeet of electronic servants. However, one can envision a single, polyfunctional, robot providing elder (and child) care, cooking meals and cleaning the house. This trend is timed between three and 10 years, reﬂecting that we’re still a long way from that ideal, but polyfunctional robots provide other beneﬁts which will see them adopted much more quickly in other environments. Today’s robots often present a single point of failure in a business process, and one which requires signiﬁcant investment to deploy and maintain. Business processes are often rearchitected to ensure that an expensive robot is fully occupied for as much time as possible, while accepting that 100% utilization is rarely possible. This rearchitecting further contributes to the cost (and risk) of deployment, while creating a production bottleneck that puts efﬁciency savings at risk if the robot does not perform as hoped. This can even threaten production itself if the robot fails in any way. These risks have slowed robot adoption outside of specialist industries with volume requirements large enough to allow for signiﬁcant redundancy. Tomorrow’s polyfunctional robots are a different breed. These robots can take on different tasks, but are also designed to ﬁt into a human-shaped world, thus removing the need for architectural changes to the process (or bolt-down infrastructure in the working environment), making for fast deployment, low risk and easy scalability. Many are humanoid, or partly humanoid, in form factor. They can thus substitute for (or be substituted by) a human worker, which increases ﬂexibility, as humans and robots can collaborate within the same working space, to the beneﬁt of both. This market is still developing, with new products being developed by existing players such as Huandi-owned Boston Dynamics, ABB and new entrants such as Tesla and Unitree. The range of products is wide: a robot dog might cost between $75,000 (Boston Dynamics’ SPOT) and $1,600 (Unitree’s Go2). Similar disparities exist in humanoid and torso form factors. Though the industry is still deciding what constitutes the minimum functionality required, and thus what price is acceptable, the development of hardware is being accompanied by huge leaps in software. Gartner, Inc. | G00815761 Page 23 of 30 This research note is restricted to the personal use of [email protected]. Training the robot to a new task now consists of showing the robot what to do. However, the polyfunctional robot isn’t just a mimic — it interprets the task and intended outcome, enabling it to modify the instructions based on the desired objective, rather than just copying what it has seen. This kind of training — combined with better motors, batteries, sensors and actuators — makes polyfunctional robots one of the most revolutionary trends on this list. Actions: Adopt a policy of “consider polyfunctionality ﬁrst” for all robot deployments, to maintain ﬂexibility and anticipate future demands by considering how the robot might need to retain value as demand changes. Start drawing up essential company policies for robot coexistence (robotology), consider how your staff will be expected to interact with robots and how they might react. For more information, see Top Strategic Technology Trends for 2025: Polyfunctional Robots. Neurological Enhancement Back to top Analysis by Sylvain Fabre, Frank Buytendijk Strategic Planning Assumptions: By 2030, 30% of knowledge workers will be enhanced by, and dependent on, technologies such as BBMIs (both employer- and self-funded) to stay relevant with the rise of AI in the workplace, up from less than 1% in 2024. By 2034, 50% of Fortune 500 companies will collect bioinformation from employee-used devices, prompting a revamp of privacy policies. Neurological enhancement improves human cognitive abilities using technologies that read and decode brain activity. It “reads” a person’s brain to provide “brain transparency” by using unidirectional brain-machine interfaces (UBMIs) or bidirectional brain-machine interfaces (BBMIs; see Glossary for deﬁnitions) and a range of other approaches. Soon, neurological enhancement will also be able to “write” to the brain, enhancing its function. Gartner, Inc. | G00815761 Page 24 of 30 This research note is restricted to the personal use of [email protected]. The UBMIs and BBMIs at the heart of neurological enhancement measure electrical activity in the user’s brain and also monitor the user’s mental state. BBMIs enable some modiﬁcation of the brain’s state based on analytics and insights. This modiﬁcation occurs in two ways: via noninvasive electrostimulation through a head-mounted wearable, or an invasive implant. Neurological enhancement’s brain-communicating capability sets it apart from other technologies that also employ wearables and implants. It’ll enable organizations to monetize: The addition of capability to the brain (e.g., in information processing, memory, learning and gaming) The extraction of information from the brain (e.g., thoughts and emotions) Much work is being done in this area. For example, Neuralink obtained FDA approval for human trials of implantable chips in May 2023, 10 a business-critical development, with the ﬁrst patient operated on in January 2024. 11 Additionally, the sums being invested in neurological enhancement are signiﬁcant. In April 2024, brain-chip maker Blackrock Neurotech received a $200-million investment from crypto ﬁrm Tether. 12 Neurological enhancement has huge potential in three main areas: Human upskilling: Enhances cognitive abilities such as memory, attention, learning and problem solving. This is vital to help people keep up with AI and for workers to stay relevant — being an “enhanced” human may become a condition of employment. Next-generation marketing: Enables brands to know, in real time, what consumers are thinking and feeling as they interact with the world. Performance: Enhances human neural capabilities to optimize outcomes in areas ranging from the prevention of industrial accidents, to learning, to healthy aging. For example, it could prevent drivers from falling asleep, personalize education for each learner and enable an aging population to stay working for longer. Use cases will evolve in this sequence: Medical research Gartner, Inc. | G00815761 Page 25 of 30 This research note is restricted to the personal use of [email protected]. Medical applications Mindfulness Workplace monitoring Personalized education Neurological enhancements for business professionals However, implementation and adoption challenges exist. They fall into several categories: Cost and technology issues: These include the high cost of early products, limited battery life, limited mobility and wireless connectivity options and the complexity of integrating different data systems. Social acceptance: More advanced functionality means more invasive and risky solutions. People will be resistant to implants because of the risk of initial surgery. This will be compounded by the need for repeat surgery because ﬁrst-generation devices will have a short life span. Social acceptance, especially for the more conspicuous form factors, may be a long way off. Sensing: Electrical activity in the brain is 3D and in depth, so measuring activity on the surface can’t capture the full “digital twin” of the brain’s electrical activity. Neither can an embedded implant. Security: UBMIs and BBMIs interface directly with the human brain. This creates security challenges and new vulnerabilities to individuals and companies. Ethics and privacy: The use of UBMIs and BBMIs raises serious ethical concerns, including issues such as altering users’ perceptions of reality, memories or even their personalities. The right to think freely must be revisited, as must the concept of privacy. 13 Actions: Set up proofs of concept of existing solutions with acceptable form factors (e.g., headphones) where you have applicable use cases (i.e., accelerated training). Do this for solutions that are already providing high returns and have high levels of acceptability. Gartner, Inc. | G00815761 Page 26 of 30 This research note is restricted to the personal use of [email protected]. Protect your organization from the issues of workers using their own consumer neurological enhancement devices by preparing now. Involve legal counsel early on to establish policies for unauthorized implantables. Keep customers and your business safe by implementing data anonymity and privacy for the collection and management of data from brain-wearables. For more information, see Top Strategic Technology Trends for 2025: Neurological Enhancement. Changes Since Last Year For 2024, Gartner identiﬁed 10 strategic technology trends (see Top Strategic Technology Trends for 2024): AI-augmented development AI trust, risk and security management Augmented connected workforce Continuous threat exposure management Democratized generative AI Industry cloud platforms Intelligent applications Machine customers Platform engineering Sustainable technology Gartner, Inc. | G00815761 Page 27 of 30 This research note is restricted to the personal use of [email protected]. Evidence 1 2024 Gartner CIO Generative AI Survey. This survey was conducted online from 30 January through 12 February 2024 to examine how CIOs are thinking about generative AI and the current role of the CIO in generative AI initiatives. This serves as an update to the 2023 Gartner CIO Generative AI Survey. In total, 83 CIOs who were members of Gartner’s Research Circle participated. Members from North America (n = 42), EMEA (n = 29), Asia/Paciﬁc (n = 7) and Latin America (n = 5) responded. (Gartner’s CIO Research Circle members include enterprise-level CIOs/CTOs, divisional CIOs/CTOs and heads of the ofﬁce of the CIO, representing a mix of industries and organization sizes.) Disclaimer: Results of this survey do not represent global ﬁndings or the market as a whole, but reﬂect the sentiments of the respondents and companies surveyed. 2 Article 27: Fundamental Rights Impact Assessment for High-Risk AI Systems, EU Artiﬁcial Intelligence Act. 3 Growing Public Concern About the Role of Artiﬁcial Intelligence in Daily Life, Pew Research Center. 4 HSBC’s Principles for the Ethical Use of Data and AI, HSBC. 5 AI Ethics at Unilever: From Policy to Process, MIT Sloan Management Review. 6 Millions of Apps in AR Glasses, XREAL. 7 A Guided Tour of Apple Vision Pro, Apple. 8 Do What You Love in Entirely New Ways, Meta. 9 Quick Answer: What Is Industrial Metaverse? 10 Elon Musk’s Neuralink Says It Has FDA Approval for Human Trials: What to Know, Washington Post. 11 ‘ People Think It’s Like the Matrix’: Neuralink’s First Patient on Having a Brain Chip, Euronews. 12 What $200 Million in Crypto Cash Means for Blackrock Neurotech, Forbes. 13 Your Brain May Not Be Private Much Longer, Vox. Gartner, Inc. | G00815761 Page 28 of 30 This research note is restricted to the personal use of [email protected]. Acronym Key and Glossary Terms Bidirectional A neural interface that enables two-way communication between Brain- a human brain and a computer or machine. It can measure the Machine user’s mental state and also inﬂuence it. Interface (BBMI) Spatial The creation of 3D (x, y, z) datasets, also called “point clouds,” to Mapping produce a 3D visualization of a (static) indoor or outdoor environment. Unidirectional A neural interface that enables one-way communication between Brain- a human brain and a computer or machine. It can measure the Machine user’s mental state and enable the user to control computers or Interface machines via thought. (UBMI) Document Revision History Top Strategic Technology Trends for 2024 - 16 October 2023 Top Strategic Technology Trends for 2023 - 17 October 2022 Top Strategic Technology Trends for 2022 - 18 October 2021 Top Strategic Technology Trends for 2021 - 19 October 2020 Gartner, Inc. | G00815761 Page 29 of 30 This research note is restricted to the personal use of [email protected]. © 2024 Gartner, Inc. and/or its afﬁliates. All rights reserved. Gartner is a registered trademark of Gartner, Inc. and its afﬁliates. This publication may not be reproduced or distributed in any form without Gartner's prior written permission. It consists of the opinions of Gartner's research organization, which should not be construed as statements of fact. While the information contained in this publication has been obtained from sources believed to be reliable, Gartner disclaims all warranties as to the accuracy, completeness or adequacy of such information. Although Gartner research may address legal and ﬁnancial issues, Gartner does not provide legal or investment advice and its research should not be construed or used as such. Your access and use of this publication are governed by Gartner's Usage Policy. Gartner prides itself on its reputation for independence and objectivity. Its research is produced independently by its research organization without input or inﬂuence from any third party. For further information, see "Guiding Principles on Independence and Objectivity." Gartner research may not be used as input into or for the training or development of generative artiﬁcial intelligence, machine learning, algorithms, software, or related technologies. Gartner, Inc. | G00815761 Page 30 of 30 This research note is restricted to the personal use of [email protected]. AI Imperatives and Risks New Frontiers of Computing Human-Machine Synergy Agentic AI Postquantum Cryptography Spatial Computing AI Governance Platforms Ambient Invisible Intelligence Polyfunctional Robots Disinformation Security Energy-Efficient Computing Neurological Enhancement Hybrid Computing Gartner, Inc. | G00815761 Page 1A of 1A This research note is restricted to the personal use of [email protected].

Gartner: Top Strategic Technology Trends for 2025 PDF

Document Details

Tags

Related

Summary

Full Transcript