Gartner unveils top technology trends at Gartner IT Symposium

ORLANDO, Fla. -- During the Gartner IT Symposium, Gartner released its top strategic technology trends for 2027. Gene Alvarez, a distinguished vice president analyst at Gartner, presented the firm's top strategic technology findings.

"The critical friction today is that AI systems operate in an architectural vacuum, generating software faster than humans can review it, creating unmanaged credential paths and binding processes to single-source vendor monopolies," said Alvarez in Gartner's release. "To prevent this systemic exposure, organizations must move from a passive observer of model capabilities to an active architect of organization, human and machine boundaries."

To come to its conclusions, Gartner uses every aspect of inputs that come in, Alvarez said. These can include venture capital funding, markets it tracks and its own hype cycles. Other methods that point to these Gartner trends include new patents, analyst submissions and insights, and innovation profiles.

These are emerging technologies, so they take time to grow – and grow at different speeds. Here are the top emerging technologies presented at the Gartner IT Symposium.

1. Large quantitative models

Large quantitative models (LQMs) can perform large-scale qualitative analyses, process complex equations and predict outcomes using mathematical equations and the laws of physics. Normal large language models (LLMs) struggle with logic because they are programmed to predict text rather than using numerical, time-series data and science data to perform forecasts. Numerical tasks can improve accuracy and enable proactive simulations for businesses.

"When you ask an LLM for an answer about math, you're asking an English major to talk about calculus because it works with text, not math," Alvarez said.

He also said businesses can use these mathematical analyses to make predictions and decisions, and that LLMs will be the interface for prompting this analysis.

2. Speech reasoning models

Speech reasoning models are voice AI systems that focus on audio and can recognize the context of a person's voice and determine next steps. For example, they can sense frustration in a customer's voice and determine when it is time to escalate a call to a person.

Using these models can let companies increase head count with no additional costs, Alvarez said in his presentation. He said these models can turn voice channels into channels with higher deflection rates because they understand what customers want.

However, these systems need time to mature and understand the complexities of inflection and tone of voice. They also introduce privacy and consent concerns, because businesses also collect people's voice patterns, which are identifiable characteristics.

Therefore, companies will need to add this information to their privacy statements for regulatory compliance, Alvarez said.

3. World models

World models use physics to enable physical AI and machines to understand and predict outcomes. They can also help autonomous agents simulate and test business strategies in advance. The example Alvarez used was driverless cars and skidding in the rain. World models can help them prepare and predict outcomes as they drive by using the laws of physics and momentum.

Gartner predicts that by 2031, businesses will be able to do 3D spatial simulations, such as a car accident, to determine possible outcomes before they happen.

Companies can pilot these models by using something that has already happened to see the predictions, evaluate whether they made a good or bad decision, and determine whether the model is ready for the organization, Alvarez said. Additionally, these models have a high compute demand.

4. AI context platforms

For AI agents to do business, they need to find information within the organization, and they do this with a context platform. Context platforms are automated AI systems that scan data and documentation across the business to help the model interpret for business decisions.

AI context platforms govern the agent's ability to work within the organization's context, which is specific to that business, Alvarez said. These platforms can integrate fragmented data, unique policies, and both structured and unstructured data, all within the context of how the organization operates.

Businesses should prioritize evaluating the interoperability capabilities of these tools.

"Think about the context player layer as a new employee. They start to learn how you do business over time," Alvarez said. "The better you feed the data to it, the stronger the context layer will get."

5. Agent coding harness

AI agent coding harnesses help write code the way the business does by managing the control plane. They are closed-loop control systems that manage the entire software development lifecycle (SDLC), relieving developers of the need to manually verify code. The written code follows company protocol, which helps keep costs under control.

Businesses that shift from writing code to using the harness to guide code development should ensure developers lead the effort, Alvarez said, because the harnesses need their knowledge to run efficiently. He also said the harness will continue to learn and get stronger, so it's important to treat it as an evolving asset.

6. AI agent identity

Agents need identity access management to govern whether they are allowed to do something, have access to specific data or interact with other agents. AI agent identity ensures the agents have the appropriate privileges and are not exploitable.

Gartner forecasts that 25% of enterprise breaches will be caused by ungoverned AI agents by 2028. In Gartner's release, Alvarez said, "Applying dynamic identity controls allows organizations to replace high-risk static credentials with dynamic, policy-controlled access."

7. Autonomous exposure remediation

With updated code to test and push into production, businesses have a hard time keeping up, creating more vulnerabilities. Autonomous exposure remediation (AER) uses both AI and automation to scan, manage and fix security vulnerabilities.

This system can also deploy patches and test code while in production. This embeds security within agentic development, and will be part of an organization's coding harness.

8. Sovereign tech

Sovereign tech gives companies control over hardware, software, databases and AI models, providing exit strategies from vendors. It also prevents vendor lock-in and reliance on foreign vendors.

By 2030, Gartner predicts more than 75% of non-U.S.-based organizations will adopt digital sovereignty strategies to limit vendor lock-in and geopolitical risks.

9. Disposable applications

Disposable applications are built for a one-time specific use and should have an expiration date when the business need has been met. The decommissioning should begin as soon as it no longer provides value to the business. These applications also have higher maintenance costs than replacement costs.

"Now, this means we need to govern applications from creation to death, looking at the entire SDLC as well as the custom software that we buy off the shelf," said Alvarez. "We need to ask: How long is it delivering business value? And when it stops, how do we decommission it?"

10. Wetware as a service

Data centers have been a major topic of conversation lately -- particularly, their electricity and water consumption. AI consumes energy at a rate that is hard to keep up with and produce.

"Think about this: Your brain runs on 20 watts. Imagine being able to run AI models at that low of an electrical consumption," Alvarez said. "Imagine biological substrates that actually do AI computing."

This is wetware as a service -- an emerging model that uses living or synthetic biological neurons to power AI.

This is currently in pilot form and will take years to evolve. Currently, there are specialty labs building data centers that use living tissue for AI computing. This method introduces new concerns for businesses, including the ethics of substrate donation and ownership.

Amanda Hetler is senior editor and writer within the Reader Engagement and Outreach team of Informa TechTarget.