Listen to the session or watch below

Whether it’s the race to find life on Mars, the campaign to outsmart killer asteroids, or the quest to make the moon a permanent home to astronauts, scientists’ efforts in space can tell us more about where humanity is headed. This subscriber-only discussion examines the progress and possibilities ahead.

Speakers: Amanda Silverman, features & investigations editor, and Robin George Andrews, award-winning science journalist and author

Recorded on March 25, 2026

Related Stories:

• America was winning the race to find Martian life. Then China jumped in.
• Inside the hunt for the most dangerous asteroid ever
• See the stunning first images from the Vera C. Rubin Observatory
• Meet the researchers testing the “Armageddon” approach to asteroid defense
• Science Writing by Robin George Andrews

We’re starting to give AI agents real autonomy, but are we prepared for what could happen next?

This subscriber-only eBook explores this and angles from experts, such as “If we continue on the current path … we are basically playing Russian roulette with humanity.”

by Grace Huckins June 12, 2025

Related Stories:

• Are we ready to hand AI agents the keys?
• MIT Technology Review Narrated: Are we ready to hand AI agents the keys?

Access all subscriber-only eBooks:

The impact of artificial intelligence extends far beyond the digital world and into our everyday lives, across the cars we drive, the appliances in our homes, and medical devices that keep people alive. More and more, product engineers are turning to AI to enhance, validate, and streamline the design of the items that furnish our worlds.

The use of AI in product engineering follows a disciplined and pragmatic trajectory. A significant majority of engineering organizations are increasing their AI investment, according to our survey, but they are doing so in a measured way. This approach reflects the priorities typical of product engineers. Errors have concrete consequences beyond abstract fears, ranging from structural failures to safety recalls and even potentially putting lives at risk. The central challenge is realizing AI’s value without compromising product integrity.

Drawing on data from a survey of 300 respondents and in-depth interviews with senior technology executives and other experts, this report examines how product engineering teams are scaling AI, what is limiting broader adoption, and which specific capabilities are shaping adoption today and, in the future, with actual or potential measurable outcomes.

Key findings from the research include:

Verification, governance, and explicit human accountability are mandatory in an environment where the outputs are physical—and the risk high. Where product engineers are using AI to directly inform physical designs, embedded systems, and manufacturing decisions that are fixed at release, product failures can lead to real-world risks that cannot be rolled back. Product engineers are therefore adopting layered AI systems with distinct trust thresholds instead of general-purpose deployments.

Predictive analytics and AI-powered simulation and validation are the top near-term investment priorities for product engineering leaders. These capabilities—selected by a majority of survey respondents—offer clear feedback loops, allowing companies to audit performance, attain regulatory approval, and prove return on investment (ROI). Building gradual trust in AI tools is imperative.

Nine in ten product engineering leaders plan to increase investment in AI in the next one to two years, but the growth is modest. The highest proportion of respondents (45%) plan to increase investment by up to 25%, while nearly a third favor a 26% to 50% boost. And just 15% plan a bigger step change—between 51% and 100%. The focus for product engineers is on optimization over innovation, with scalable proof points and near-term ROI the dominant approach to AI adoption, as opposed to multi-year transformation.

Sustainability and product quality are top measurable outcomes for AI in product engineering. These outcomes, visible to customers, regulators, and investors, are prioritized over competitive metrics like time to-market and innovation—rated of medium importance—and internal operational gains like cost reduction and workforce satisfaction, at the bottom. What matters most are real-world signals like defect rates and emissions profiles rather than internal engineering dashboards.

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. 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.

In the race to adopt and show value from AI, enterprises are moving faster than ever to deploy agentic AI as copilots, assistants, and autonomous task-runners. In late 2025, nearly two-thirds of companies were experimenting with AI agents, while 88% were using AI in at least one business function, up from 78% in 2024, according to McKinsey’s annual AI report. Yet, while early pilots often succeed, only one in 10 companies actually scaled their AI agents.

One major issue: AI agents are only as effective as the data foundation supporting them. Experts argue that most companies are seeing delays in implementing AI, not because of shortcomings in the models, but because they lack data architectures that deliver business context to be reliably used by humans and agents.

Companies need to be ready with the right data architecture, and the next few months — years, at most — will be critical, says Irfan Khan, president and chief product officer of SAP Data & Analytics.

“The only prediction anybody can reliably make is that we don’t know what’s going to happen in the years, months — or even weeks — ahead with AI,” he says. “To be able to get quick wins right now, you need to adopt an AI mindset and … ground your AI models with reliable data.”

While data has always been important for business, it will be even more so in the age of AI. The capabilities of agentic AI will be set more by the soundness of enterprise data architecture and governance, and less by the evolution of the models. To scale the technology, businesses need to adopt a modern data infrastructure that delivers context along with the data.

More business context, not necessarily more data

Traditional views often conflate structured data with high value, and unstructured data with less value. However, AI complicates that distinction. High-value data for agents is defined less by format and more by business context. Data for critical business functions — such as supply-chain operations and financial planning — is context dependent. While fine-grained, high-volume data, such as IoT, logs, and telemetry, can yield value, but only when delivered with business context.

For that reason, the real risk for agentic AI is not lack of data, but lack of grounding, says Khan.

“Anything that is business contextual will, by definition, give you greater value and greater levels of reliability of the business outcome,” he says. “It’s not as simple as saying high-value data is structured data and low-value data is where you have lots of repetition — both can have huge value in the right hands, and that’s what’s different about AI.”

Context can be derived through integration with software, on-site analysis and enrichment, or through the governance pipeline. Data lacking those qualities will likely be untrusted — one reason why two-thirds of business leaders do not fully trust their data, according to the Institute for Data and Enterprise AI (IDEA). The resulting “trust debt” has held back businesses in their quest for AI readiness. Overcoming that lack of trust requires shared definitions, semantic consistency, and reliable operational context to align data with business meaning.

Data sprawl demands a semantic, business-aware layer

Over the past decade, the most important shift in enterprise data architecture has been the separation of compute and storage, cloud-scale flexibility, says Khan. Yet, that separation and move to cloud also created sprawl, with data housed in multiple clouds, data lakes, warehouses, and a multitude of SaaS applications.

As companies move to AI, that sprawl does not go away. In fact, the problem is growing with more than two-thirds of companies citing data siloes as a top challenge in adopting AI, with more than half of enterprises struggling with 1,000 data sources or more. While the last era was about laying the foundation on which to build software-as-a-service — separating compute and storage and building lakes — the next era is about delivering the right data to autonomous AI agents tasked with various business functions.

“Probably the biggest innovation that occurred in data management was the separation of compute and store,” Khan says. “But what’s really making a distinction now is the way that we harmonize the data and harvest the value of the data across multiple sources of content.”

To do that requires a semantic or knowledge layer that supports multiple platforms, encodes business rules and relationships, provides a business-contextual and governed view of data, and allows humans and agents to access the data in the appropriate ways. But legacy data architectures cannot power the autonomous AI systems of the future, consultancy Deloitte stated in its State of AI in the Enterprise report. Only four in 10 companies believe their data management process is ready for AI, and that’s down from 43% the previous year, suggesting that as companies explore AI deployment, they are realizing their infrastructure’s shortcomings.

Agentic AI does not replace SaaS

Some investors and technologists speculate that AI agents will make SaaS applications obsolete. Khan strongly disagrees. Over the past 15 years, value has steadily moved up the stack, from on-premises infrastructure to infrastructure as a service (IaaS) to platform as a service (PaaS) to SaaS. Agentic AI is simply the next layer. Agentic AI will have its own layer to access the data and interact with the business logic. The value rises up the stack, but nothing below disappears, he says.

“SaaS doesn’t go away,” he says. “It just means SaaS and these agents will cooperate with one another. Companies are not going to throw away their entire general ledger and replace it with an agent. What’s the agent going to do? It doesn’t know anything without business context and business processing.”

In this emerging model, the software stack is being reshaped so that applications and data provide governed context within which AI can act effectively. SaaS applications remain the systems of record, while the semantic layer becomes the business-context source of truth. AI agents become a new engagement layer, orchestrating across systems, and both humans and agents become “first-class citizens” in how they access business logic, he says.

Critically, agents cannot directly connect to every operational system. “If we’re saying agents are going to take over the world … you can’t have an agent talking to every operational backend system,” Khan warns. “It just doesn’t work that way.”

This further elevates the importance of a semantic or business-fabric layer.

Where to start

Most enterprises need to begin where their data already lives — in platforms like Snowflake, Databricks, Google BigQuery, or an existing SAP environment. Khan says that’s normal, but warns against rebuilding old patterns of vendor lock-in.

He suggests that companies prioritize the data that matters most by focusing on preserving and providing business context to operational and application data. Companies should also invest early in governance and semantics by defining shared policies, access rules, and semantic models before scaling pilots. Finally, businesses should prioritize openness and fabric-style interoperability rather than forcing all data into one stack.

Khan cautions against aiming for full automation too early. “There is a new brave opportunity to really engage in the agentic and AI world,” Khan says, “Fully automating [critical business processes] is maybe a stretch, because there’s going to be a lot of extra oversight necessary.” Early wins will likely come from less-critical processes and from agents that work off fresh, stateful data rather than stale dashboards, he adds. As AI begins to deliver value and adoption increases, leaders must decide how to reinvest those gains to drive top-line efficiency or enter new markets.

Register for “The Fabric of Data & AI” virtual event on March 24, 2026. Hear insights from executives and thought leaders who are shaping the future of data and AI.

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.

Loudoun County, Virginia, once known for its pastoral scenery and proximity to Washington, DC, has earned a more modern reputation in recent years: The area has the highest concentration of data centers on the planet.

Ten years ago, these facilities powered email and e-commerce. Today, thanks to the meteoric rise in demand for AI-infused everything, local utility Dominion Energy is working hard to keep pace with surging power demands. The pressure is so acute that Dulles International Airport is constructing the largest airport solar installation in the country, a highly visible bid to bolster the region’s power mix.

Data center campuses like Loudoun’s are cropping up across the country to accommodate an insatiable appetite for AI. But this buildout comes at an enormous cost. In the US alone, data centers consumed roughly 4% of national electricity in 2024. Projections suggest that figure could stretch to 12% by 2028. To put this in perspective, a single 100-megawatt data center consumes roughly as much electricity as 80,000 American homes. Data centers being built today are gearing up for gigawatt scale, enough to power a mid-sized city.

For enterprise leaders, energy costs associated with AI and data infrastructure are quickly becoming both a budget concern and a potential bottleneck on growth. Meeting this moment calls for a capability most organizations are only beginning to develop: energy intelligence. The emerging discipline refers to understanding where, when, and why energy is consumed, and using that insight to optimize operations and control costs.

These efforts stand to address both immediate financial pressures and longer-term reputational risks, as communities like Loudoun County grow increasingly concerned about the energy demands associated with nearby data center development.

In December 2025, MIT Technology Review Insights conducted a survey of 300 executives to understand how companies are thinking about energy intelligence today, as well as where they’re anticipating challenges in the future.

Here are five of our most notable findings:

• Energy intelligence is becoming a universal business priority. One hundred percent of executives surveyed expect the ability to measure and strategically manage power consumption to become an important business metric in the next two years.
  
• AI workloads are already driving measurable cost increases, and the surge is just beginning. Two-thirds of executives (68%) report their companies have faced energy cost increases of 10% or more in the past 12 months due to AI and data workloads. Nearly all respondents (97%) anticipate their organization’s AI-related energy consumption will increase over the next 12-18 months.
  
• Mounting costs are the top energy-related threat to AI innovation. Half of executives (51%) rank rising costs as the single greatest energy-related risk to their digital and AI initiatives. Most companies currently tracking and attempting to optimize data center energy consumption are motivated by cost management.
  
• Organizations are responding through infrastructure optimization and energy-efficient partnerships. To address mounting energy demands, three in four leaders (74%) are optimizing existing infrastructure, while 69% are partnering with energy-efficient cloud and storage providers. More than half are also implementing AI workload scheduling (61%) and investing in more efficient hardware (56%).
  
• Closing the measurement gap is the next frontier. Most enterprises still lack the granular data needed for true energy intelligence. This gap is especially pronounced for companies relying on third-party cloud providers and managed services for their data compute and storage needs, where 71% say rising consumption-based costs originate, yet energy metrics are often opaque.

Download the full 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. 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.

When Tony Fadell started working on the iPod, usability often trumped security. The result was an iterative process. Every time someone would find a security weakness or a way to hack the device, the development group would iterate to add measures and fix the issues. Yet, flaws would frequently be found, and the secure design of the product became a moving target.

But when it came to designing a device specifically for security purposes, there could be no iterative process after rolling it out: Security had to be the number one priority. 

“As you develop these things, you’re a victim of your own development speed,” says Fadell, who developed Ledger Stax, a signing device for securing digital assets, and is now a board member at digital asset security firm Ledger. “If you introduced these features and functions without the proper review, and now customers are demanding security, you’ll realize that you should have designed it differently from the start, and it’s very hard to undo what you’ve already done.”

A critical aspect of designing secure technology, however, must be ease of use too. Without it, it is all too simple for users to make a mistake or use an unsafe workaround that undermines device protections. Think a post-it stuck to a monitor or some variation of “123456” or “admin” for passwords.

With digital asset security devices like signers—more commonly called “wallets”—such errors could lead to seriously detrimental outcomes. If, for example, a user’s private key falls into the wrong hands, bad actors can use it to steal their digital assets. Estimates suggest that around 20% of all Bitcoin—worth around $355 billion—are inaccessible to owners. One of the reasons for this is likely because they lost their private keys.

In the past, crypto devices have been notoriously difficult to use. As cryptocurrency becomes ever more popular, valuable, and mainstream—attracting greater attention from criminals as the stakes rise—designers and engineers are prioritizing both security and usability when developing digital asset devices, drawing on in-depth research to iterate.

The three components of security

Strong security models for devices like signers, which are used to secure blockchain transactions,  require three major components. First, a secure operating system. Second, a secure element to bind the software to the hardware. And third, a secure user interface. Each of which need to be frequently tested by researchers and white hat hackers to simulate real-world attacks and improve product resilience and usability.

The first two elements focus on securing the device software and hardware. Secure software has always been a problem, but one that has improved over the last decade, as security architectures and processes have been refined. Meanwhile, hardware security components have become widely available—from trusted platform modules on computers to secure enclaves in smartphones—allowing digital information to essentially be locked to a device.

For crypto signers, hardware must provide encryption capabilities. And the security of the software must be frequently tested. Ledger, for example, has a secure OS and a Secure Element that handles encryption primitives, and a secure display that prevents device takeover.

Security and usability working hand in hand

Asset recovery is a major consideration when designing signers. If recovery options are not easy to use, an owner could lose access. But if recovery processes are not secure enough, attackers could exploit the system. With SIM swapping attacks, for example, attackers can tap into a mobile communications channel used for account recovery and “recover” a victim’s password to steal their assets.

In the digital-asset ecosystem, the creation of the seed phrase, a sequence of 12 to 24 words that could act as a passphrase for wallets is an example of improving usability and security. Known more formally as Bitcoin Improvement Proposal 39 (BIP-39), the approach gives users a master password to unlock their hierarchical deterministic (HD) wallets. 

There is a lot of creative tension between the security team and the UX team that happens to achieve the proper balance between convenience and safety, Fadell says, referring to Ledger’s security research team, the Donjon. “We mock things up, we prototype things from a UX UI perspective, we walk through it, then we walk the Donjon team through it,” Fadell explains. “We push back and forth to find the absolute optimal solution to balance the two.” 

Through the research the Donjon team has conducted, Ledger designed its Recovery Key—an NFC-based physical card to back up your 24 words—to be both user-friendly and secure. “What we did, as a first in the industry, was include an NFC card,” says Fadell. “Instead of only writing it down, you can also have an NFC card called a Recovery Key. You can have multiple Recovery Keys and store them in a lockbox, a safety deposit box, or give them to someone you trust for safekeeping.”

A number of government initiatives are working to regulate this balance between security and usability. This includes the US Cybersecurity and Infrastructure Security Agency’s Secure by Design, which aims to build cybersecurity into the design and manufacture of technology products. And the UK’s National Cyber Security Centre’s Software Security Code of Practice, which outlines security principles expected of all organizations that develop or sell software. 

Enterprise security presents distinct challenges

Embedding usability and security into devices for companies adds further complexity as businesses need features such as multi-signature capabilities to protect against single points of failure, whether from external attacks or internal bad actors. 

Security design can take these requirements into account, with secure governance using multiple signatures (multisig), hardware security modules (HSMs) for key storage, trusted display systems, and other usable security capabilities.

These technologies are critically important for companies who have roles in the blockchain ecosystem. Failure to establish robust security measures can have dire consequences. In 2024, for example, unknown cybercriminals made off with more than $300 million worth of assets from DMM Bitcoin, leading the Japanese cryptocurrency platform to close six months later. Japan’s Financial Services Agency discovered severe risk management issues, including inadequate oversight, lack of independent audits, and poor security practices.

For companies, allowing a multi-stage process that involves a required number of stakeholders is critical, says Fadell. “It’s making sure that the attack vector is not just one person, and so you need to support multiple people with multiple factors on all of their devices as well,” he says. “It gets to be a real combinatoric problem.”

R&D to stay one step ahead 

To keep up with requirements and offer strong security with improved visibility, crypto firms need to invest in research and development, Fadell says. Attack labs, such as Ledger Donjon, can conduct real-world testing on specific enterprise security requirements and create scenarios to educate both management and workers of the potential threats. 

Such research and development can support device designers and engineers in their never-ending mission to balance security measures with usability so that digital asset devices can support users to safeguard their digital assets in a constantly evolving crypto and cyber landscape.

Learn more about how to secure digital assets in the Ledger Academy.

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.