Realizing value with AI inference at scale and in production

Training an AI model to predict equipment failures is an engineering achievement. But it’s not until prediction meets action—the moment that model successfully flags a malfunctioning machine—that true business transformation occurs. One technical milestone lives in a proof-of-concept deck; the other meaningfully contributes to the bottom line.

Craig Partridge, senior director worldwide of Digital Next Advisory at HPE, believes “the true value of AI lies in inference”. Inference is where AI earns its keep. It’s the operational layer that puts all that training to use in real-world workflows. Partridge elaborates, “The phrase we use for this is ‘trusted AI inferencing at scale and in production,’” he says. “That’s where we think the biggest return on AI investments will come from.”

Getting to that point is difficult. Christian Reichenbach, worldwide digital advisor at HPE, points to findings from the company’s recent survey of 1,775 IT leaders: While nearly a quarter (22%) of organizations have now operationalized AI—up from 15% the previous year—the majority remain stuck in experimentation.

Reaching the next stage requires a three-part approach: establishing trust as an operating principle, ensuring data-centric execution, and cultivating IT leadership capable of scaling AI successfully.

Trust as a prerequisite for scalable, high-stakes AI

Trusted inference means users can actually rely on the answers they’re getting from AI systems. This is important for applications like generating marketing copy and deploying customer service chatbots, but it’s absolutely critical for higher-stakes scenarios—say, a robot assisting during surgeries or an autonomous vehicle navigating crowded streets.

Whatever the use case, establishing trust will require doubling down on data quality; first and foremost, inferencing outcomes must be built on reliable foundations. This reality informs one of Partridge’s go-to mantras: “Bad data in equals bad inferencing out.”

Reichenbach cites a real-world example of what happens when data quality falls short—the rise of unreliable AI-generated content, including hallucinations, that clogs workflows and forces employees to spend significant time fact-checking. “When things go wrong, trust goes down, productivity gains are not reached, and the outcome we’re  looking for is not achieved,” he says.

On the other hand, when trust is properly engineered into inference systems, efficiency and productivity gains can increase. Take a network operations team tasked with troubleshooting configurations. With a trusted inferencing engine, that unit gains a reliable copilot that can deliver faster, more accurate, custom-tailored recommendations—”a 24/7 member of the team they didn’t have before,” says Partridge.

The shift to data-centric thinking and rise of the AI factory

In the first AI wave, companies rushed to hire data scientists and many viewed sophisticated, trillion-parameter models as the primary goal. But today, as organizations move to turn early pilots into real, measurable outcomes, the focus has shifted toward data engineering and architecture.

“Over the past five years, what’s become more meaningful is breaking down data silos, accessing data streams, and quickly unlocking value,” says Reichenbach. It’s an evolution happening alongside the rise of the AI factory—the always-on production line where data moves through pipelines and feedback loops to generate continuous intelligence.

This shift reflects an evolution from model-centric to data-centric thinking, and with it comes a new set of strategic considerations. “It comes down to two things: How much of the intelligence–the model itself–is truly yours? And how much of the input–the data–is uniquely yours, from your customers, operations, or market?” says Reichenbach.

These two central questions inform everything from platform direction and operating models to engineering roles and trust and security considerations. To help clients map their answers—and translate them into actionable strategies—Partridge breaks down HPE’s four-quadrant AI factory implication matrix (see figure):

Source: HPE, 2025

• Run: Accessing an external, pretrained model via an interface or API; organizations don’t own the model or the data. Implementation requires strong security and governance. It also requires establishing a center of excellence that makes and communicates decisions about AI usage.
• RAG (retrieval augmented generation): Using external, pre-trained models combined with a company’s proprietary data to create unique insights. Implementation focuses on connecting data streams to inferencing capabilities that provide rapid, integrated access to full-stack AI platforms.
• Riches: Training custom models on data that resides in the enterprise for unique differentiation opportunities and insights. Implementation requires scalable, energy-efficient environments, and often high-performance systems.
• Regulate: Leveraging custom models trained on external data, requiring the same scalable setup as Riches, but with added focus on legal and regulatory compliance for handling sensitive, non-owned data with extreme caution.

Importantly, these quadrants are not mutually exclusive. Partridge notes that most organizations—including HPE itself—operate across many of the quadrants. “We build our own models to help understand how networks operate,” he says. “We then deploy that intelligence into our products, so that our end customer gets the chance to deliver in what we call the ‘Run’ quadrant. So for them, it’s not their data; it’s not their model. They’re just adding that capability inside their organization.”

IT’s moment to scale—and lead

The second part of Partridge’s catchphrase about inferencing—”at scale”— speaks to a primary tension in enterprise AI: what works for a handful of use cases often breaks when applied across an entire organization.

“There’s value in experimentation and kicking ideas around,” he says. “But if you want to really see the benefits of AI, it needs to be something that everybody can engage in and that solves for many different use cases.”

In Partridge’s view, the challenge of turning boutique pilots into organization-wide systems is uniquely suited to the IT function’s core competencies—and it’s a leadership opportunity the function can’t afford to sit out. “IT takes things that are small-scale and implements the discipline required to run them at scale,” he says. “So, IT organizations really need to lean into this debate.”

For IT teams content to linger on the sidelines, history offers a cautionary tale from the last major infrastructure shift: enterprise migration to the cloud. Many IT departments sat out decision-making during the early cloud adoption wave a decade ago, while business units independently deployed cloud services. This led to fragmented systems, redundant spending, and security gaps that took years to untangle.

The same dynamic threatens to repeat with AI, as different teams experiment with tools and models outside IT’s purview. This phenomenon—sometimes called shadow AI—describes environments where pilots proliferate without oversight or governance. Partridge believes that most organizations are already operating in the “Run” quadrant in some capacity, as employees will use AI tools whether or not they’re officially authorized to.

Rather than shut down experimentation, it is now IT’s mandate to bring structure to it. And enterprises must architect a data platform strategy that brings together enterprise data with guardrails, governance framework, and accessibility to feed AI. Also, it’s critical to keep standardizing infrastructure (such as private cloud AI platforms), protecting data integrity, and safeguarding brand trust, all while enabling the speed and flexibility that AI applications demand. These are the requirements for reaching the final milestone: AI that’s truly in production.

For teams on the path to that goal, Reichenbach distills what success requires. “It comes down to knowing where you play: When to Run external models smarter, when to apply RAG to make them more informed, where to invest to unlock Riches from your own data and models, and when to Regulate what you don’t control,” says Reichenbach. “The winners will be those who bring clarity to all quadrants and align technology ambition with governance and value creation.”

For more, register to watch MIT Technology Review’s EmTech AI Salon, featuring HPE.

This content was produced by Insights, the custom content arm of MIT Technology Review. It was not written by MIT Technology Review’s editorial staff. It was researched, designed, and written by human writers, editors, analysts, and illustrators. This includes the writing of surveys and collection of data for surveys. AI tools that may have been used were limited to secondary production processes that passed thorough human review.

Networking for AI: Building the foundation for real-time intelligence

The Ryder Cup is an almost-century-old tournament pitting Europe against the United States in an elite showcase of golf skill and strategy. At the 2025 event, nearly a quarter of a million spectators gathered to watch three days of fierce competition on the fairways.

From a technology and logistics perspective, pulling off an event of this scale is no easy feat. The Ryder Cup’s infrastructure must accommodate the tens of thousands of network users who flood the venue (this year, at Bethpage Black in Farmingdale, New York) every day.

To manage this IT complexity, Ryder Cup engaged technology partner HPE to create a central hub for its operations. The solution centered around a platform where tournament staff could access data visualization supporting operational decision-making. This dashboard, which leveraged a high-performance network and private-cloud environment, aggregated and distilled insights from diverse real-time data feeds.

It was a glimpse into what AI-ready networking looks like at scale—a real-world stress test with implications for everything from event management to enterprise operations. While models and data readiness get the lion’s share of boardroom attention and media hype, networking is a critical third leg of successful AI implementation, explains Jon Green, CTO of HPE Networking. “Disconnected AI doesn’t get you very much; you need a way to get data into it and out of it for both training and inference,” he says.

As businesses move toward distributed, real-time AI applications, tomorrow’s networks will need to parse even more massive volumes of information at ever more lightning-fast speeds. What played out on the greens at Bethpage Black represents a lesson being learned across industries: Inference-ready networks are a make-or-break factor for turning AI’s promise into real-world performance.

Making a network AI inference-ready

More than half of organizations are still struggling to operationalize their data pipelines. In a recent HPE cross-industry survey of 1,775  IT leaders, 45% said they could run real-time data pushes and pulls for innovation. It’s a noticeable change over last year’s numbers (just 7% reported having such capabilities in 2024), but there’s still work to be done to connect data collection with real-time decision-making.

The network may hold the key to further narrowing that gap. Part of the solution will likely come down to infrastructure design. While traditional enterprise networks are engineered to handle the predictable flow of business applications—email, browsers, file sharing, etc.—they’re not designed to field the dynamic, high-volume data movement required by AI workloads. Inferencing in particular depends on shuttling vast datasets between multiple GPUs with supercomputer-like precision.

“There’s an ability to play fast and loose with a standard, off-the-shelf enterprise network,” says Green. “Few will notice if an email platform is half a second slower than it might’ve been. But with AI transaction processing, the entire job is gated by the last calculation taking place. So it becomes really noticeable if you’ve got any loss or congestion.”

Networks built for AI, therefore, must operate with a different set of performance characteristics, including ultra-low latency, lossless throughput, specialized equipment, and adaptability at scale. One of these differences is AI’s distributed nature, which affects the seamless flow of data.

The Ryder Cup was a vivid demonstration of this new class of networking in action. During the event, a Connected Intelligence Center was put in place to ingest data from ticket scans, weather reports, GPS-tracked golf carts, concession and merchandise sales, spectator and consumer queues, and network performance. Additionally, 67 AI-enabled cameras were positioned throughout the course. Inputs were analyzed through an operational intelligence dashboard and provided staff with an instantaneous view of activity across the grounds.

“The tournament is really complex from a networking perspective, because you have many big open areas that aren’t uniformly packed with people,” explains Green. “People tend to follow the action. So in certain areas, it’s really dense with lots of people and devices, while other areas are completely empty.”

To handle that variability, engineers built out a two-tiered architecture. Across the sprawling venue, more than 650 WiFi 6E access points, 170 network switches, and 25 user experience sensors worked together to maintain continuous connectivity and feed a private cloud AI cluster for live analytics. The front-end layer connected cameras, sensors, and access points to capture live video and movement data, while a back-end layer—located within a temporary on-site data center—linked GPUs and servers in a high-speed, low-latency configuration that effectively served as the system’s brain. Together, the setup enabled both rapid on-the-ground responses and data collection that could inform future operational planning. “AI models also were available to the team which could process video of the shots taken and help determine, from the footage, which ones were the most interesting,” says Green.

Physical AI and the return of on-prem intelligence

If time is of the essence for event management, it’s even more critical in contexts where safety is on the line—for instance a self-driving car making a split-second decision to accelerate or brake.

In planning for the rise of physical AI, where applications move off screens and onto factory floors and city streets, a growing number of enterprises are rethinking their architectures. Instead of sending the data to centralized clouds for inference, some are deploying edge-based AI clusters that process information closer to where it is generated. Data-intensive training may still occur in the cloud, but inferencing happens on-site.

This hybrid approach is fueling a wave of operational repatriation, as workloads once relegated to the cloud return to on-premises infrastructure for enhanced speed, security, sovereignty, and cost reasons. “We’ve had an out-migration of IT into the cloud in recent years, but physical AI is one of the use cases that we believe will bring a lot of that back on-prem,” predicts Green, giving the example of an AI-infused factory floor, where a round-trip of sensor data to the cloud would be too slow to safely control automated machinery. “By the time processing happens in the cloud, the machine has already moved,” he explains.

There’s data to back up Green’s projection: research from Enterprise Research Group shows that 84% of respondents are reevaluating application deployment strategies due to the growth of AI. Market forecasts also reflect this shift. According to IDC, the AI market for infrastructure is expected to reach $758 billion by 2029.

AI for networking and the future of self-driving infrastructure

The relationship between networking and AI is circular: Modern networks make AI at scale possible, but AI is also helping make networks smarter and more capable.

“Networks are some of the most data-rich systems in any organization,” says Green. “That makes them a perfect use case for AI. We can analyze millions of configuration states across thousands of customer environments and learn what actually improves performance or stability.”

At HPE for example, which has one of the largest network telemetry repositories in the world, AI models analyze anonymized data collected from billions of connected devices to identify trends and refine behavior over time. The platform processes more than a trillion telemetry points each day, which means it can continuously learn from real-world conditions.

The concept broadly known as AIOps (or AI-driven IT operations) is changing how enterprise networks are managed across industries. Today, AI surfaces insights as recommendations that administrators can choose to apply with a single click. Tomorrow, those same systems might automatically test and deploy low-risk changes themselves.

That long-term vision, Green notes, is referred to as a “self-driving network”—one that handles the repetitive, error-prone tasks that have historically plagued IT teams. “AI isn’t coming for the network engineer’s job, but it will eliminate the tedious stuff that slows them down,” he says. “You’ll be able to say, ‘Please go configure 130 switches to solve this issue,’ and the system will handle it. When a port gets stuck or someone plugs a connector in the wrong direction, AI can detect it—and in many cases, fix it automatically.”

Digital initiatives now depend on how effectively information moves. Whether coordinating a live event or streamlining a supply chain, the performance of the network increasingly defines the performance of the business. Building that foundation today will separate those who pilot from those who scale AI.

For more, register to watch MIT Technology Review’s EmTech AI Salon, featuring HPE.

This content was produced by Insights, the custom content arm of MIT Technology Review. It was not written by MIT Technology Review’s editorial staff. It was researched, designed, and written by human writers, editors, analysts, and illustrators. This includes the writing of surveys and collection of data for surveys. AI tools that may have been used were limited to secondary production processes that passed thorough human review.

Improving VMware migration workflows with agentic AI

For years, many chief information officers (CIOs) looked at VMware-to-cloud migrations with a wary pragmatism. Manually mapping dependencies and rewriting legacy apps mid-flight was not an enticing, low-lift proposition for enterprise IT teams.

But the calculus for such decisions has changed dramatically in a short period of time. Following recent VMware licensing changes, organizations are seeing greater uncertainty around the platform’s future. At the same time, cloud-native innovation is accelerating. According to the CNCF’s 2024 Annual Survey, 89% of organizations have already adopted at least some cloud-native techniques, and the share of companies reporting nearly all development and deployment as cloud-native grew sharply from 2023 to 2024 (20% to 24%). And market research firm IDC reports that cloud providers have become top strategic partners for generative AI initiatives.

This is all happening amid escalating pressure to innovate faster and more cost-effectively to meet the demands of an AI-first future. As enterprises prepare for that inevitability, they are facing compute demands that are difficult, if not prohibitively expensive, to maintain exclusively on-premises.

Download the full article.

This content was produced by Insights, the custom content arm of MIT Technology Review. It was not written by MIT Technology Review’s editorial staff.

This content was researched, designed, and written by human writers, editors, analysts, and illustrators. This includes the writing of surveys and collection of data for surveys. AI tools that may have been used were limited to secondary production processes that passed thorough human review.

Reimagining cybersecurity in the era of AI and quantum

AI and quantum technologies are dramatically reconfiguring how cybersecurity functions, redefining the speed and scale with which digital defenders and their adversaries can operate.

The weaponization of AI tools for cyberattacks is already proving a worthy opponent to current defenses. From reconnaissance to ransomware, cybercriminals can automate attacks faster than ever before with AI. This includes using generative AI to create social engineering attacks at scale, churning out tens of thousands of tailored phishing emails in seconds, or accessing widely available voice cloning software capable of bypassing security defenses for as little as a few dollars. And now, agentic AI raises the stakes by introducing autonomous systems that can reason, act, and adapt like human adversaries.

But AI isn’t the only force shaping the threat landscape. Quantum computing has the potential to seriously undermine current encryption standards if developed unchecked. Quantum algorithms can solve the mathematical problems underlying most modern cryptography, particularly public-key systems like RSA and Elliptic Curve, widely used for secure online communication, digital signatures, and cryptocurrency.

“We know quantum is coming. Once it does, it will force a change in how we secure data across everything, including governments, telecoms, and financial systems,” says Peter Bailey, senior vice president and general manager of Cisco’s security business.

“Most organizations are understandably focused on the immediacy of AI threats,” says Bailey. “Quantum might sound like science fiction, but those scenarios are coming faster than many realize. It’s critical to start investing now in defenses that can withstand both AI and quantum attacks.”

Critical to this defense is a zero trust approach to cybersecurity, which assumes no user or device can be inherently trusted. By enforcing continuous verification, zero trust enables constant monitoring and ensures that any attempts to exploit vulnerabilities are quickly detected and addressed in real time. This approach is technology-agnostic and creates a resilient framework even in the face of an ever-changing threat landscape.

Putting up AI defenses 

AI is lowering the barrier to entry for cyberattacks, enabling hackers even with limited skills or resources to infiltrate, manipulate, and exploit the slightest digital vulnerability.

Nearly three-quarters (74%) of cybersecurity professionals say AI-enabled threats are already having a significant impact on their organization, and 90% anticipate such threats in the next one to two years. 

“AI-powered adversaries have advanced techniques and operate at machine speed,” says Bailey. “The only way to keep pace is to use AI to automate response and defend at machine speed.”

To do this, Bailey says, organizations must modernize systems, platforms, and security operations to automate threat detection and response—processes that have previously relied on human rule-writing and reaction times. These systems must adapt dynamically as environments evolve and criminal tactics change.

At the same time, companies must strengthen the security of their AI models and data to reduce exposure to manipulation from AI-enabled malware. Such risks could include, for instance, prompt injections, where a malicious user crafts a prompt to manipulate an AI model into performing unintended actions, bypassing its original instructions and safeguards.

Agentic AI further ups the ante, with hackers able to use AI agents to automate attacks and make tactical decisions without constant human oversight. “Agentic AI has the potential to collapse the cost of the kill chain,” says Bailey. “That means everyday cybercriminals could start executing campaigns that today only well-funded espionage operations can afford.”

Organizations, in turn, are exploring how AI agents can help them stay ahead. Nearly 40% of companies expect agentic AI to augment or assist teams over the next 12 months, especially in cybersecurity, according to Cisco’s 2025 AI Readiness Index. Use cases include AI agents trained on telemetry, which can identify anomalies or signals from machine data too disparate and unstructured to be deciphered by humans. 

Calculating the quantum threat

As many cybersecurity teams focus on the very real AI-driven threat, quantum is waiting on the sidelines. Almost three-quarters (73%) of US organizations surveyed by KPMG say they believe it is only a matter of time before cybercriminals are using quantum to decrypt and disrupt today’s cybersecurity protocols. And yet, the majority (81%) also admit they could do more to ensure that their data remains secure.

Companies are right to be concerned. Threat actors are already carrying out harvest now, decrypt later attacks, stockpiling sensitive encrypted data to crack once quantum technology matures. Examples include state-sponsored actors intercepting government communications and cybercriminal networks storing encrypted internet traffic or financial records. 

Large technology companies are among the first to roll out quantum defenses. For example, Apple is using cryptography protocol PQ3 to defend against harvest now, decrypt later attacks on its iMessage platform. Google is testing post-quantum cryptography (PQC)—which is resistant to attacks from both quantum and classical computers—in its Chrome browser. And Cisco “has made significant investments in quantum-proofing our software and infrastructure,” says Bailey. “You’ll see more enterprises and governments taking similar steps over the next 18 to 24 months,” he adds. 

As regulations like the US Quantum Computing Cybersecurity Preparedness Act lay out requirements for mitigating against quantum threats, including standardized PQC algorithms by the National Institute of Standards and Technology, a wider range of organizations will start preparing their own quantum defenses. 

For organizations beginning that journey, Bailey outlines two key actions. First, establish visibility. “Understand what data you have and where it lives,” he says. “Take inventory, assess sensitivity, and review your encryption keys, rotating out any that are weak or outdated.”

Second, plan for migration. “Next, assess what it will take to support post-quantum algorithms across your infrastructure. That means addressing not just the technology, but also the process and people implications,” Bailey says.

Adopting proactive defense 

Ultimately, the foundation for building resilience against both AI and quantum is a zero trust approach, says Bailey. By embedding zero trust access controls across users, devices, business applications, networks, and clouds, this approach grants only the minimum access required to complete a task and enables continuous monitoring. It can also minimize the attack surface by confining a potential threat to an isolated zone, preventing it from accessing other critical systems.

Into this zero trust architecture, organizations can integrate specific measures to defend against AI and quantum risks. For instance, quantum-immune cryptography and AI-powered analytics and security tools can be used to identify complex attack patterns and automate real-time responses. 

“Zero trust slows down attacks and builds resilience,” Bailey says. “It ensures that even if a breach occurs, the crown jewels stay protected and operations can recover quickly.”

Ultimately, companies should not wait for threats to emerge and evolve. They must get ahead now. “This isn’t a what-if scenario; it’s a when,” says Bailey. “Organizations that invest early will be the ones setting the pace, not scrambling to catch up.”

This content was produced by Insights, the custom content arm of MIT Technology Review. It was not written by MIT Technology Review’s editorial staff. It was researched, designed, and written by human writers, editors, analysts, and illustrators. This includes the writing of surveys and collection of data for surveys. AI tools that may have been used were limited to secondary production processes that passed thorough human review.

Leveraging the clinician’s expertise with agentic AI

Building a high performance data and AI organization (2nd edition)

Four years is a lifetime when it comes to artificial intelligence. Since the first edition of this study was published in 2021, AI’s capabilities have been advancing at speed, and the advances have not slowed since generative AI’s breakthrough. For example, multimodality— the ability to process information not only as text but also as audio, video, and other unstructured formats—is becoming a common feature of AI models. AI’s capacity to reason and act autonomously has also grown, and organizations are now starting to work with AI agents that can do just that.

Amid all the change, there remains a constant: the quality of an AI model’s outputs is only ever as good as the data
that feeds it. Data management technologies and practices have also been advancing, but the second edition of this study suggests that most organizations are not leveraging those fast enough to keep up with AI’s development. As a result of that and other hindrances, relatively few organizations are delivering the desired business results from their AI strategy. No more than 2% of senior executives we surveyed rate their organizations highly in terms of delivering results from AI.

To determine the extent to which organizational data performance has improved as generative AI and other AI advances have taken hold, MIT Technology Review Insights surveyed 800 senior data and technology executives. We also conducted in-depth interviews with 15 technology and business leaders.

Key findings from the report include the following:

• Few data teams are keeping pace with AI. Organizations are doing no better today at delivering on data strategy than in pre-generative AI days. Among those surveyed in 2025, 12% are self-assessed data “high achievers” compared with 13% in 2021. Shortages of skilled talent remain a constraint, but teams also struggle with accessing fresh data, tracing lineage, and dealing with security complexity—important requirements for AI success.

• Partly as a result, AI is not fully firing yet. There are even fewer “high achievers” when it comes to AI. Just 2% of respondents rate their organizations’ AI performance highly today in terms of delivering measurable business results. In fact, most are still struggling to scale generative AI. While two thirds have deployed it, only 7% have done so widely.

Download the report.

This content was produced by Insights, the custom content arm of MIT Technology Review. It was not written by MIT Technology Review’s editorial staff. It was researched, designed, and written by human writers, editors, analysts, and illustrators. AI tools that may have been used were limited to secondary production processes that passed thorough human review.

Finding return on AI investments across industries

The market is officially three years post ChatGPT and many of the pundit bylines have shifted to using terms like “bubble” to suggest reasons behind generative AI not realizing material returns outside a handful of technology suppliers. 

In September, the MIT NANDA report made waves because the soundbite every author and influencer picked up on was that 95% of all AI pilots failed to scale or deliver clear and measurable ROI. McKinsey earlier published a similar trend indicating that agentic AI would be the way forward to achieve huge operational benefits for enterprises. At The Wall Street Journal’s Technology Council Summit, AI technology leaders recommended CIOs stop worrying about AI’s return on investment because measuring gains is difficult and if they were to try, the measurements would be wrong. 

This places technology leaders in a precarious position–robust tech stacks already sustain their business operations, so what is the upside to introducing new technology? 

For decades, deployment strategies have followed a consistent cadence where tech operators avoid destabilizing business-critical workflows to swap out individual components in tech stacks. For example, a better or cheaper technology is not meaningful if it puts your disaster recovery at risk. 

While the price might increase when a new buyer takes over mature middleware, the cost of losing part of your enterprise data because you are mid-way through transitioning your enterprise to a new technology is way more severe than paying a higher price for a stable technology that you’ve run your business on for 20 years.

So, how do enterprises get a return on investing in the latest tech transformation?

First principle of AI: Your data is your value

Most of the articles about AI data relate to engineering tasks to ensure that an AI model infers against business data in repositories that represent past and present business realities. 

However, one of the most widely-deployed use cases in enterprise AI begins with prompting an AI model by uploading file attachments into the model. This step narrows an AI model’s range to the content of the uploaded files, accelerating accurate response times and reducing the number of prompts required to get the best answer. 

This tactic relies upon sending your proprietary business data into an AI model, so there are two important considerations to take in parallel with data preparation: first, governing your system for appropriate confidentiality; and second, developing a deliberate negotiation strategy with the model vendors, who cannot advance their frontier models without getting access to non-public data, like your business’ data. 

Recently, Anthropic and OpenAI completed massive deals with enterprise data platforms and owners because there is not enough high-value primary data publicly available on the internet. 

Most enterprises would automatically prioritize confidentiality of their data and design business workflows to maintain trade secrets. From an economic value point of view, especially considering how costly every model API call really is, exchanging selective access to your data for services or price offsets may be the right strategy. Rather than approaching model purchase/onboarding as a typical supplier/procurement exercise, think through the potential to realize mutual benefits in advancing your suppliers’ model and your business adoption of the model in tandem.

Second principle of AI: Boring by design

According to Information is Beautiful, in 2024 alone, 182 new generative AI models were introduced to the market. When GPT5 came into the market in 2025, many of the models from 12 to 24 months prior were rendered unavailable until subscription customers threatened to cancel. Their previously stable AI workflows were built on models that no longer worked. Their tech providers thought the customers would be excited about the newest models and did not realize the premium that business workflows place on stability. Video gamers are happy to upgrade their custom builds throughout the entire lifespan of the system components in their gaming rigs, and will upgrade the entire system just to play a newly released title. 

However, behavior does not translate to business run rate operations. While many employees may use the latest models for document processing or generating content, back-office operations can’t sustain swapping a tech stack three times a week to keep up with the latest model drops. The back-office work is boring by design.

The most successful AI deployments have focused on deploying AI on business problems unique to their business, often running in the background to accelerate or augment mundane but mandated tasks. Relieving legal or expense audits from having to manually cross check individual reports but putting the final decision in a humans’ responsibility zone combines the best of both. 

The important point is that none of these tasks require constant updates to the latest model to deliver that value. This is also an area where abstracting your business workflows from using direct model APIs can offer additional long-term stability while maintaining options to update or upgrade the underlying engines at the pace of your business.

Third principle of AI: Mini-van economics

The best way to avoid upside-down economics is to design systems to align to the users rather than vendor specs and benchmarks. 

Too many businesses continue to fall into the trap of buying new gear or new cloud service types based on new supplier-led benchmarks rather than starting their AI journey from what their business can consume, at what pace, on the capabilities they have deployed today. 

While Ferrari marketing is effective and those automobiles are truly magnificent, they drive the same speed through school zones and lack ample trunk space for groceries. Keep in mind that every remote server and model touched by a user layers on the costs and design for frugality by reconfiguring workflows to minimize spending on third-party services. 

Too many companies have found that their customer support AI workflows add millions of dollars of operational run rate costs and end up adding more development time and cost to update the implementation for OpEx predictability. Meanwhile, the companies that decided that a system running at the pace a human can read—less than 50 tokens per second—were able to successfully deploy scaled-out AI applications with minimal additional overhead.

There are so many aspects of this new automation technology to unpack—the best guidance is to start practical, design for independence in underlying technology components to keep from disrupting stable applications long term, and to leverage the fact that AI technology makes your business data valuable to the advancement of your tech suppliers’ goals.

This content was produced by Intel. It was not written by MIT Technology Review’s editorial staff.

Redefining data engineering in the age of AI

As organizations weave AI into more of their operations, senior executives are realizing data engineers hold a central role in bringing these initiatives to life. After all, AI only delivers when you have large amounts of reliable and well-managed, high-quality data. Indeed, this report finds that data engineers play a pivotal role in their organizations as enablers of AI. And in so doing, they are integral to the overall success of the business.

According to the results of a survey of 400 senior data and technology executives, conducted by MIT Technology Review Insights, data engineers have become influential in areas that extend well beyond their traditional remit as pipeline managers. The technology is also changing how data engineers work, with the balance of their time shifting from core data management tasks toward AI-specific activities.

As their influence grows, so do the challenges data engineers face. A major one is dealing with greater complexity, as more advanced AI models elevate the importance of managing unstructured data and real-time pipelines. Another challenge is managing expanding workloads; data engineers are being asked to do more today than ever before, and that’s not likely to change.

Key findings from the report include the following:

• Data engineers are integral to the business. This is the view of 72% of the surveyed technology leaders—and 86% of those in the survey’s biggest organizations, where AI maturity is greatest. It is a view held especially strongly among executives in financial services and manufacturing companies.
• AI is changing everything data engineers do. The share of time data engineers spend each day on AI projects has nearly doubled in the past two years, from an average of 19% in 2023 to 37% in 2025, according to our survey. Respondents expect this figure to continue rising to an average of 61% in two years’ time. This is also contributing to bigger data engineer workloads; most respondents (77%) see these growing increasingly heavy.

Download the report.

This content was produced by Insights, the custom content arm of MIT Technology Review. It was not written by MIT Technology Review’s editorial staff.

This content was researched, designed, and written by human writers, editors, analysts, and illustrators. This includes the writing of surveys and collection of data for surveys. AI tools that may have been used were limited to secondary production processes that passed thorough human review.