Dan Huttenlocher doesn’t like walls. This isn’t so much an aesthetic preference as it is a practical concern. Walls divide people and define spaces. They restrict movement. They discourage exchange. And they’re a pain to move if your needs change, especially when they’re stuffed with cables, ducts, and other infrastructural accessories.

Mr. Huttenlocher is certain his needs will change. As dean of Cornell Tech, a closely watched collaboration in New York City between Cornell University and the Technion-Israel Institute of Technology, Mr. Huttenlocher is overseeing the creation of an institution dedicated to technological innovation, academic experimentation, and the kind of serial flexibility those two principles require.

\”My goal as the dean is to create an environment where everything can be repurposed,\” Mr. Huttenlocher says.

He and his team are in the tenuous middle stages of planning and building exactly that: the chameleon campus, a space where interchangeability permeates everything. As Cathy Dove, Cornell Tech’s founding vice president, puts it, \”We want to embody the principle of iteration.\”

This attempt at building in nimbleness is a hedge against the hissing pace of technological change. Cornell Tech expects to open its first buildings in 2017 and its last two decades later. Yet the campus is being planned now by people who know they cannot imagine how the intervening years will change the way we interact with the digital world, maybe even with each other. In shunning any semblance of permanence, Mr. Huttenlocher hopes to answer the central question of this daring and expensive endeavor: \”How do you do something that’s technologically advanced that isn’t immediately technologically dated?\”

To ask it another way, how do you create a new institution in an age where everything—office design, intelligent infrastructure, cloud computing, classroom technology—presents some opportunity to break with the past? What do you build? What do you wire? What kind of interactions do you encourage? Some institutions might create committees to try to anticipate specific changes. Cornell Tech is determined to do the opposite. Those responsible for building the campus of the future won’t pretend to know what the future holds. They only hope they’re building something malleable enough to handle it.

A Campus Unlike Any Other

Cornell and Technion joined forces in December 2011 when New York City selected them to build an applied-science graduate school on Manhattan’s Roosevelt Island. The announcement followed an intense and highly publicized competition among the Cornell-Technion partnership and six other higher-education groups, all lured by the city’s offer to donate land and contribute to capital costs. The Cornell-Technion partnership won based on the strength of its coalition and the grandiosity of its plans: two million square feet, 2,000 students, 280 professors, $2-billion. Michael R. Bloomberg, who was then New York City\’s mayor, called it \”far and away the boldest and most ambitious\” proposal received.

Even after winning the competition, Cornell Tech isn’t easing off its lofty promises. Its website boasts, \”We’re creating a campus on Roosevelt Island that’s unlike any ever created in higher education.\” At present that campus is a demolition zone containing the carcass of what was once Goldwater Memorial Hospital. But behind the scenes Cornell Tech’s soaring rhetoric is beginning to take the shape of hard commitments.

This summer Cornell Tech finalized the construction design for its first academic building. It has also made crucial decisions about its IT infrastructure and begun experimenting with classroom technologies. Earlier this year the institution released a video detailing the first phase of the project, which will include a residential tower, a \”colocation building\” for industry partners, and the first academic building. Rubber, in other words, is meeting road, and it starts with what Mr. Huttenlocher calls the \”hardscape.\”

That includes the floor plan for the first academic building, which was finalized just two months ago. The second, third, and fourth stories of the five-level structure are stunningly undefined, dominated by large, uninterrupted spaces. Classrooms? Faculty offices? The building will have little of the former and none of the latter. Instead there are \”office zones,\” which will be filled with workstations; those seeking some form of enclosure can enter a \”huddle room,\” \”swing space, \”collab\” room, or \”hub lounge.\” The entrepreneurial patois, conspicuous as it sounds, reflects a real attempt to break down traditional academic boundaries.

\”We want this building to support and encourage collaboration across very different groups of people who might normally be siloed in different places across a university,\” says Mr. Huttenlocher. \”That’s a goal that we don’t see changing over a 75-year time frame.\”

Already, though, Mr. Huttenlocher says he’s getting pushback from faculty members on the absence of offices. \”My goal as the dean is to create an environment where it will be very open-plan like this. Does every faculty member agree with that view? No,\” Mr. Huttenlocher says. He adds, \”I don’t know whether I’ll be successful in that cultural change or not.\”

Smart Buildings

Less contentious but equally intriguing is the technology integrated into the building design. David Wilts, an IT specialist with the design firm Arup and a consultant on the project, says Cornell Tech is attempting to create what he calls the \”holy grail\” of building-management systems, the so-called smart building.

\”The idea of a smart building has been around theoretically for decades,\” Mr. Wilts says. \”It was first really talked about in great detail in the 70s, but only in the last five years has the technology price point dropped low enough.\”

The broad concept is to have a structure where all the building-maintenance functions—heating, cooling, ventilation, lighting, fire safety, etc.—are fully automated and share data with one another on a common network. For example, the light fixtures in Cornell Tech’s first academic building will have sensors capable of reading room occupancy, and the light-sensor data will feed into an IT network that also controls the heating, cooling, and scheduling systems. If a room still isn’t occupied, say, 10 minutes after the time for which it was reserved, the lights will dim, the AC will shut off, and the room will be reclassified as \”available\” on a room-scheduling app.

Mr. Wilts believes that kind of interchange on a network spanning the campus will open eyes. \”People talk about getting every building system onto a network. That doesn’t really accomplish anything,\” says Mr. Wilts. \”What really accomplishes things is getting the data to actually be exchanged between all of these different systems and applications.\”

The exact specifications of the building-management system are not yet certain. Unlike the physical plant, the fixtures that make up this smart-building network will need replacing about every decade, the team at Cornell Tech anticipates. How then to create something that works in 2017 but is flexible enough to accommodate the technologies of future decades? Mr. Wilts says the key \”is to use as many open standards and protocols as possible. If you’re using open protocols and open languages, devices then become interchangeable.\” He points to the scheduling system, which was designed to accommodate a specific calendar and email program but has been recalibrated to become \”generic.\”

The long-term effectiveness of this network will rely on the agility and reliability of the campus IT infrastructure. And on that front, the school is making perhaps its most notable gamble: It will not build a data center. \”We fundamentally do not have a data center,\” says Scott Yoest, director of IT. \”We don’t even think that way.\”

Data centers, which house large numbers of servers, cost money. They’re bulky. They suck down energy. They’ve also been a staple of college campuses for decades, providing the computing power and local storage necessary to accommodate research and daily use. \”It’s been amazing for me after 25 years in IT higher ed to think of a campus without a data center,\” says Mr. Yoest. \”I stumbled on it for about the first three months.\”

The idea of building a campus without a data center was Mr. Huttenlocher’s. He and Mr. Yoest had what the latter calls a \”heart to heart\” about the decision in 2012. The move has obvious appeal given the quantum leaps in cloud computing, much of it driven by commercial vendors. Why build your own data storage when industry leaders can do the heavy lifting for you? It’s a notion higher-ed IT professionals say they’ve been mulling over for a while.

\”I guarantee there is no one in higher ed more reliable than Google,\” says Sharif Nijim, enterprise application architect at the University of Notre Dame. \”You can’t compete with that scale. You can’t compete with that salary paradigm. You’re not going to attract that kind of talent. Take advantage of the fact that other people are landing that talent.\”

Mr. Nijim says if he were starting a campus from scratch today, he’d forgo a data center. \”You’re not saddled with all the detritus that’s built up over time,\” he says. \”That’s your biggest advantage.\” But the decision comes with risks, especially at an applied-sciences institution where data demands should be higher than at, say, a liberal-arts campus. The embedded assumption is that cloud computing will improve fast enough to accommodate a growing graduate institution.

Finding What Works

In its embrace of technological progress, the data-center decision encapsulates much of the Cornell Tech mind-set. It also exposes the risk of flirting with what Mr. Huttenlocher calls the \”bleeding edge.\” In other words, it’s great to be at the technological forefront, until the technology doesn’t work.

And at present, there is a whole lot of technology that doesn’t work for Cornell Tech. Its temporary campus inside Google’s New York headquarters houses all sorts of experimental playthings. One room has three different videoconferencing systems, setting up a sort of natural selection based on whichever gets the most use. A device called the BeamPro—best described as remote-controlled Segway meets videoconferencing—allows professors to roam the campus hallways virtually even if in reality they’re halfway around the world. And classrooms are outfitted with a technology that enables laptops to wirelessly transmit to a common flat-screen television. At least that’s what it’s supposed to do. \”The first round of experiments has been close to a dismal failure,\” Mr. Huttenlocher says.

In 2015, Cornell Tech will begin testing a prototype classroom to be used in the first academic building. A year after that, the IT team will do what Mr. Yoest calls a \”deep dive\” on audiovisual technology. The gadgetry may be the most immediately appealing part of what Cornell Tech is doing. After all, what would the self-proclaimed campus of \”the next century\” be without toys? But to make it to the next century with its reputation intact, the institution’s planners will have to avoid investing too much in the kind of proprietary technology that tends to expire quickly.

There is, in that sense, a sort of Zen to the planning process. Committing to flexibility means committing to nothing. Embracing technology means not buying too much of it. \”We certainly know we’re going to need classroom technologies,\” says Ms. Dove, the vice president. \”But what will classroom technology look like in three years? Nobody knows.\”

In the meantime, Cornell Tech will keep experimenting, careful not to make too many commitments or erect too many barriers. This summer it will create a new testing ground by turning what is now the capital-planning room into another teaching space. The models and blueprints will move to another office while the existing room joins with the one adjacent to it.

\”That wall is already coming down,\” Mr. Huttenlocher says, gesturing toward a red wall lined with splashy architectural renderings. \”If I could, I’d take a sledgehammer to it myself.\”