For most of a generation, the story of connectivity was a story about arrival. The frontier was the schoolhouse door, and the goal was to get a fast enough signal across it. That frontier has largely been settled. Fiber now reaches nearly every school in the country. And so the interesting question has quietly moved indoors, from the property line to the wiring in the walls.

Two shifts are redrawing the map inside the building. Passive Optical LAN changes how the network is built. Fault-managed power, now recognized in code as Class 4, changes how power is delivered. Together they suggest a different way of thinking about a school: not as a data network with an electrical system bolted alongside it, but as one converged system. That idea is worth taking seriously. It is also worth taking skeptically, because the most useful thing anyone can do with a promising technology is describe not only when it works, but when it doesn’t.

Rethinking the network

Passive Optical LAN, or POLAN, rests on an elegant idea. Instead of stacking powered switches in a telecom closet on every floor, a single optical line terminal at the core sends light down a strand of glass to passive splitters, which fan the signal out to classrooms. Passive is the operative word. Those splitters need no power, no cooling, and no software updates, and fiber can reach many kilometers from that central point. A whole campus can run on electronics housed in one location.

The mechanism behind the savings is easy to state and hard to argue with: a converged optical design reduces the number of powered network closets a district must build, staff, cool, and eventually replace. Fewer distributed active electronics means less recurring cost, less heat, and fewer points of failure to chase. That is a structural property of the architecture, not a claim about any one building.

The dollar figures are where honesty matters most. Published cost studies and district reports generally point toward savings, but most originate with vendors or the industry’s trade association and should be read with that interest in mind. The reasonable synthesis is directional, not precise: optical LAN can lower first-cost and recurring cost in the right building, largely by eliminating distributed active electronics, but the exact figure depends entirely on the building, the labor market, and the replacement timeline. Anyone quoting you a single confident percentage is selling something.

Rethinking the power

Fault-managed power, formally Class 4, is the first genuinely new category of electrical power the code has recognized in decades. This is not a marketing term. It was written into the 2023 National Electrical Code as new Article 726, and it is governed by safety standards UL 1400-1 for systems and UL 1400-2 for cable. When the National Fire Protection Association gives a technology its own article, it signals an intent to build on it.

What makes Class 4 notable is how it stays safe. Traditional wiring limits danger by capping how much power the source can push. Fault-managed power instead sends energy in continuously monitored packets and watches the circuit in real time, shutting off within milliseconds when it detects trouble. The conductor is, in effect, touch-safe. That safety unlocks reach ordinary low-voltage cabling cannot match: fault-managed power can deliver hundreds of watts over more than a kilometer, well beyond the roughly 90 watts and 100 meters that Power over Ethernet allows.

That touch-safe promise is not a marketing claim. It is earned by the transmitter’s ability to detect and interrupt specific fault conditions before harmful energy can transfer. Industry summaries of the UL 1400-1 and Article 726 requirements generally group these into six fault conditions a Class 4 system must catch:

  1. Short circuit / overcurrent — a sudden, abnormal surge in current.
  2. Ground fault (line-to-ground) — current leaking to ground, a shock and fire risk.
  3. Line-to-line fault — an unintended connection between conductors.
  4. Arc fault — the characteristic electrical signature of an arc.
  5. Human contact (touch fault) — the impedance change of a person contacting a conductor.
  6. Cable damage or faulty wiring — changes in the line’s electrical characteristics from a cut, crush, or degraded run.

A note of precision: the standards do not publish these as a single tidy numbered list, and authoritative sources group them slightly differently (some collapse line-to-line and ground faults, others add component failure of the transmitter or receiver itself). The point is not the exact count but the principle: the system’s safety case rests on continuously testing every energy packet against a defined set of fault signatures and refusing to transmit when any one is present.

A related efficiency claim circulates in the market and is worth citing carefully rather than adopting at face value. A manufacturer study has reported that a Power-over-Ethernet lighting design can use roughly 30 percent fewer fixtures than a conventional layout, in part because networked luminaires can be zoned and controlled more granularly. That figure is a manufacturer’s result, not an independent finding, and belongs in the same directional, read-with-interest category as any other vendor number. The general and better-established point stands on its own: because touch-safe PoE and Class 4 cabling are treated more like signal wiring than branch-circuit power, they can often be installed by the technology cabling team rather than requiring licensed electricians on every run, which changes the labor math for a small district.

The convergence argument in this piece is not just an engineering preference. The National Electrical Code is being restructured around it. The 2026 NEC (NFPA 70) retires the old term low voltage in favor of limited energy, defined as the equipment and cables of an end-to-end system that are either power-restricted or able to limit or shut down the source to mitigate shock and fire hazards. For the first time, interior communications and fiber are folded out of their long-separate Chapter 8 and into a unified Chapter 7 framework, with new general requirements (Article 720) and a single consolidated cable article (Article 722).

Under that umbrella, the cabling for the following systems is now treated as one converged family:

  • Class 2 circuits — thermostats, doorbells, most security sensors, control wiring (Article 725).
  • Class 3 circuits — some nurse-call, industrial control, higher-power signaling (Article 725).
  • Class 4 fault-managed power (FMPS) — the fault-managed and digital-electricity systems at the center of this essay (Article 726).
  • Communications circuits — voice and data, structured Category cabling (formerly Article 800-series).
  • Optical fiber — the POLAN backbone and its feeders (Article 770).
  • Power-limited fire alarm (PLFA) — initiating devices and notification appliances (Article 760).
  • Coax, network-powered broadband, and antenna systems — CATV/broadband and antenna wiring (Articles 820, 830, 810).

Two caveats in keeping with the honesty of this piece. First, the full “limited energy” consolidation is a feature of the 2026 code; Class 4 itself first appeared in the 2023 edition. Second, the NEC is adopted state by state, and many jurisdictions are still on the 2020 or 2023 cycle. So these systems “fall under the converged limited-energy category” as a matter of code direction, but not yet as a matter of uniform enforcement everywhere. Check your adopting authority before relying on it.

The deeper signal: native DC, microgrids, and backup

Here is the part of the story that runs deeper than cabling, and it is the strongest signal of where this is all heading. Fault-managed power is, at bottom, a direct-current distribution technology. That single fact changes the argument, because it lets a building stop fighting its own power.

Consider how much of a modern school already runs on DC. LED lighting, every PoE device, sensors, laptops and tablets, phones, displays, the electronics inside almost everything, all of it is natively DC. In a conventional building, grid alternating current is converted to DC at each of those devices, hundreds of little conversions, each one shedding a few percent as waste heat. Now add on-site generation and storage: solar panels and batteries are also natively DC. In an AC building, their DC output is converted up to AC to join the panel, then back down to DC at the load, a DC-to-AC-to-DC round trip that exists only because the distribution system speaks the wrong language. A Class 4 transmitter, by contrast, is designed to accept input directly from the utility grid, from solar and wind, or from a DC microgrid paired with battery storage, and to distribute that power as managed DC. Every conversion you remove is efficiency you keep.

Convert once, distribute DC, invert back only where needed

This naturally raises the design question: should a building simply convert everything to DC at the utility entrance and back to AC only at the exceptions? Directionally yes, but not literally. The honest architecture is not one giant rectifier at the service entrance. It is a hybrid AC/DC bus, where the two current types coexist and power electronics form the managed boundary between them. The standards community has largely converged on this, with a building DC bus around 380 volts as the efficiency sweet spot: high enough to move real power over copper, low enough to stay manageable. You convert incoming AC to that DC bus once, tie solar and storage into it with no round trip at all, and invert back to AC only for the loads that genuinely want it.

Because the realistic building has layers, not a single conversion:

Layer Likely form Why
Service entrance AC in, or a main AC/DC rectifier The grid is AC; motors and legacy loads tap here
Solar + storage tie-in Native DC No round-trip conversion at all
Building / campus bus ~380 VDC Efficient distribution, the standards sweet spot
Managed distribution to endpoints Class 4 / FMPS Touch-safe, fault-managed DC over long runs
Edge devices Low-voltage DC (PoE, 48 V, etc.) Lighting, sensors, network, displays

The honest exceptions matter. The big rotating machinery in a school, HVAC compressors, large air handlers, pumps, elevators, is dominated by AC induction motors chosen for predictable behavior under imperfect power. You can run them from DC, but usually by adding a drive per motor, and modern variable-frequency drives already rectify to DC internally anyway. Sometimes the cheaper, more robust choice is simply to leave a lean dedicated AC branch for the mechanical room. There is also a real safety wrinkle: protecting higher-voltage DC is harder than AC, because a DC arc has no natural zero-crossing to help extinguish it, so disconnects and arc mitigation at 380 VDC are more specialized. This is precisely where Class 4 earns its place, its packet-based, fault-managed design is the DC protection scheme, but at the managed-distribution layer, not as the main service rectifier.

So the defensible version of the thesis is not a fully DC building. It is this: converge toward DC wherever loads and sources are already DC, keep a lean AC branch for rotating machinery and legacy loads, and let power electronics, with Class 4 as the managed DC layer, form the intelligent boundary between the two. Convert once for the majority of the building; invert back only where the load truly needs it.

Honesty requires care with the magnitude, because this is exactly where enthusiasts overreach. At hyperscale data-center scale, where nearly every load is DC and the conversions are many, native-DC studies cite savings on the order of 10 to 15 percent. But rigorous, peer-reviewed studies at ordinary building-distribution scale find a far more modest advantage, often only 1 to 2 percent, while a hardware DC-microgrid prototype reported around 15 percent in its own test conditions. The wide spread is the real finding: the size of the prize depends almost entirely on how DC-native your loads and your sources already are. A school with LED lighting, PoE everywhere, and rooftop solar captures far more of it than a conventional building retrofitted halfway. Treat the efficiency gain as real in direction and highly variable in size, not as a fixed number.

The more compelling half of this signal is not efficiency at all. It is resilience. Because Class 4 centralizes power at a managed transmitter, that transmitter becomes the natural marriage point for a battery bank and an on-site source, in other words, a microgrid with a built-in ability to island from the grid during an outage. For a rural school, which so often doubles as a community’s emergency shelter and warming center, that capability is the payoff that dollars-per-kilowatt-hour miss entirely. A converged, DC-capable building can keep its network, its emergency lighting, its communications, and its safety systems alive when the grid goes down, from a single centrally managed store of energy. In our region, where outages are a fact of winter life, that is not a luxury feature. It is the point.

I want to be clear that this is the most visionary and least proven part of the argument. The hybrid DC-native building is still emergent, the distribution voltages and interconnection standards are still settling, and the largest efficiency claims assume a building whose loads and sources are already DC. This is a direction of travel and a strong one, not a finished blueprint. But the signal is unmistakable: as loads go DC, as solar and storage go DC, and as the code makes touch-safe DC distribution legal and standardized, the case for pouring AC through a building whose devices all quietly convert it back grows harder to defend.

A signal to the optical equipment makers

If the converged building is where this is headed, then fault-managed power is a signal, and the manufacturers of optical networking equipment should read it. Here is the argument, offered as an informed inference rather than a settled fact.

Today’s optical network terminals, the endpoints that sit in the classroom and hand off Ethernet to devices, were largely designed for a data-first world. Their power budgets and their Power-over-Ethernet capabilities are often modest, sized for a phone and an access point. But in a converged building, that same endpoint is being asked to feed lighting, cameras, displays, sensors, and access control. The endpoint is becoming a local power distribution point, not just a data drop.

That shift argues for a new generation of optical network terminals built deliberately for smart buildings: endpoints with larger aggregate power budgets, more powered ports, and the intelligence to negotiate and deliver the full range of PoE standards, port by port. An ONT should be able to serve a 15-watt sensor on one port and a 90-watt display on the next, sensing each device’s class and modulating delivery accordingly, all fed upstream by fault-managed power that carries the energy the long distance from a central node. In effect, the optical endpoint becomes the graceful handoff between Class 4 power on the backbone and Class 2 PoE at the edge.

This is where I think the market should go, and I would encourage the optical equipment makers to build toward it. It is also fair to name the caveat: this is a direction of travel, not a shipping product line, and standards for how fault-managed power and optical endpoints interoperate cleanly are still maturing. Districts should ask for a roadmap, not assume one exists.

Where this doesn’t pencil out

An argument that cannot describe its own limits is a pitch, not an analysis. Here is where a district should slow down or say no.

We do not yet have a built proof of concept. This is the honest heart of the matter. The convergence case rests today on standards, design logic, and projections, not on a measured demonstration site. What communities need now is a district willing to build a converged pilot and an independent, ideally peer-reviewed, evaluation of what it actually costs, saves, and requires to operate. Until then, we are reasoning from design and code, not from measured outcomes, and we should say so.

Vendor concentration is a real dependency. Optical LAN, in particular, tends to be a single-vendor ecosystem: the terminals, the endpoints, and the management software often do not interoperate across manufacturers. For a district making a twenty-year infrastructure decision, that is a strategic risk worth naming out loud. Ask hard questions about open standards, second-source availability, and what happens to support if a vendor exits the market.

Class 4 is young, and the workforce is younger. Article 726 dates only to 2023. The installed base is thin, and in rural regions especially, contractors who have actually worked with fault-managed power may be scarce. The advantage that technology staff can pull touch-safe cabling is real, but it also raises questions of liability, inspection, and code familiarity when a novel power system is energized in an occupied school. Do not treat immaturity as a detail.

Centralization concentrates risk. Collapsing distributed closets into a single network-and-power node is efficient, but it also creates a more consequential point of failure. That is manageable with redundancy, backup power, and disciplined design, but it is a cost and a responsibility, not a free byproduct of convergence. The same centralization that enables a microgrid also means the node must be protected and backed up with real discipline.

And sometimes the old way is simply right. A building already slated for replacement, a district with ample in-house electrical labor, or a site with no local Class 4 expertise may be better served by conventional switching and standard power. Convergence is a strong default in new construction and deep renovation. It is a weaker case as a mid-life retrofit.

Why it still matters

With those caveats honestly on the table, the core idea holds. POLAN and fault-managed power work at different scales, fiber reaching many kilometers to carry data across a district, Class 4 reaching a shorter one-to-two-kilometer range to energize the edge, but within the footprint of a school or campus they can share a single pathway and a single origin. That convergence is what makes a genuinely modern building possible: lighting, sensors, access control, and communications fed from one place, over one backbone, with power and data managed as a single system, and increasingly as native DC drawn from on-site solar and storage.

The moment to act on this is specific. A district opening its walls for a consolidation or a major renovation faces exactly the decision point where convergence deserves serious study, because the backbone is being chosen and a converged design is far cheaper to build in now than to retrofit later. For that district, this is a foundation worth studying seriously, and if the projections hold, the return would be measured in decades, not budget cycles. Whether they hold is precisely what a demonstration site would tell us.

The policy point survives the skepticism too, though it needs stating precisely. Our public funding programs still describe school infrastructure in a vocabulary written for a pre-convergence world. Naming converged technologies in something like the E-Rate Eligible Services List will not, and should not, guarantee funding. But eligibility frameworks that cannot even name a technology cannot evaluate it on cost-effectiveness or subject it to competitive bidding. The first step is not funding. It is recognition, followed by honest scrutiny. That is a standard convergence should have to meet, not evade.

The frontier moved indoors. Our maps should follow, with eyes open.

What these ideas are based on

Standards and code (independent)

  • National Electrical Code (NFPA 70), 2023 edition, Article 726, Class 4 Fault-Managed Power Systems, and Article 100 definition of Fault-Managed Power.
  • National Electrical Code (NFPA 70), 2026 edition, restructuring of limited-energy systems: Article 720 (general requirements), Article 722 (consolidated cables), and the Article 100 definition of Limited Energy System; consolidation of interior communications and optical fiber into Chapter 7.
  • UL 1400-1 (fault-managed power systems / equipment) and UL 1400-2 (Class 4 cable) safety requirements.
  • Telecommunications Industry Association (TIA), “A Closer Look at Class 4 Fault-Managed Power Systems in Smart Buildings,” on Class 4 transmitter inputs from grid AC, renewable DC, and DC microgrids with battery storage.
  • EMerge Alliance hybrid AC/DC and DC-microgrid building standards, on ~380 VDC distribution and single-conversion architectures.
  • IEEE 802.3 Power over Ethernet standards, for the ~90W / 100m comparison baseline and the PoE classes an optical endpoint would negotiate.

Native-DC and microgrid efficiency (tiered by scale)

  • Peer-reviewed building- and distribution-scale studies (e.g., PLOS One, 2025; Frontiers in Energy Research, 2024) finding a modest ~1–2% native-DC advantage at ordinary distribution scale, and a hardware DC-microgrid prototype reporting ~15% under its own test conditions. Cited to show the range, not a single figure.
  • Data-center-scale sources (Lawrence Berkeley National Laboratory demonstrations; Schneider Electric; IEEE Spectrum, 2026) citing ~10–15% energy savings and ~20% downtime reduction where nearly all loads are DC. Applicable to high-conversion environments, not directly to a typical school.
  • Cisco-led industry white paper, “Fault Managed Power: Evolution of DC Power Distribution,” on Class 4 across smart buildings, telecom, and DC microgrids. Industry-authored; read as directional.

Industry and vendor material (directional, read with interest in mind)

  • Manufacturer study reporting a ~30% reduction in fixtures for a Power-over-Ethernet lighting design. Cited as a vendor result, not an independent finding. [Add exact manufacturer and title.]
  • Association for Passive Optical LAN (APOLAN) cost-comparison studies, cited as directional rather than definitive because the association exists to promote the technology.
  • Other manufacturer technology briefs and case studies on optical LAN and fault-managed power, useful for mechanism and range but subject to selection bias.

A note on sourcing: much of the published cost and efficiency evidence for these technologies originates with vendors or their trade association, and is flagged as directional here. Native-DC efficiency figures are tiered by scale, because data-center results do not transfer to a typical school. This op-ed argues from independent standards and code, and from the engineering properties of the two technologies, rather than from any single built or peer-reviewed deployment. The limited-energy system list reflects the 2026 NEC restructuring; Class 4 first appeared in the 2023 edition, and code adoption varies by state. The six-fault list is an industry synthesis of the UL 1400-1 / Article 726 requirements, not a verbatim clause. The arguments about hybrid AC/DC distribution, native-DC microgrids, and next-generation optical network terminals are the author’s forward-looking inferences. Readers weighing a real decision should seek independent references and, ideally, a demonstration site with an independent evaluation.