Haitz's Law¶

The empirical LED regularity that light output per package rises ~20x per decade while cost per lumen falls ~10x — a log-linear trajectory arising because half a dozen loss terms improve multiplicatively, used as a roadmap and a below-trend diagnostic until efficiencies hit their thermodynamic ceilings.

Version: v5 · 2026-09-19 · History · 3 corrections
Domain-specific #: 478
Domain group: Professional & Organizational Practice
Origin domain: Human-Computer Interaction (HCI)
Subdomain: Optoelectronics and Solid State Lighting → Human-Computer Interaction (HCI)

Core Idea¶

Haitz's law (Roland Haitz, ~1999) is the empirical regularity that light output per LED package rises by roughly 20x per decade while cost per lumen falls roughly 10x — equivalently, a doubling in lumens-per-package roughly every 28 months and a halving of cost per lumen roughly every three years. It is not one advance but the multiplicative compounding of several loss-reduction axes (internal quantum efficiency, light extraction, current handling, thermal management, phosphor conversion, yield), yielding a log-linear trajectory. It is the solid-state-lighting analogue of Moore's law.

How would you explain it like I'm…

Brighter-Cheaper Light Trend

LED lights are tiny lights in lamps and screens. For many years, each little LED got brighter and brighter, and the light it made got cheaper and cheaper, at a steady speed. That steady pattern is called Haitz's Law, but it can't go on forever, because a light can only get so good.

The LED Getting-Better Trend

Haitz's Law describes a trend that Roland Haitz spotted in LEDs. Over about ten years, the light coming out of one LED package grew roughly 20 times, while the cost for each bit of light dropped to about a tenth. This happened because engineers improved many things at once: turning electricity into light better, letting more light escape, handling more current, removing heat, and making them with fewer mistakes. Those improvements multiplied together. But it isn't a forever rule of nature: there's a limit to how efficient LEDs can get, so the trend slows down.

LED Brightness-Cost Trend

Haitz's Law, named for Roland Haitz who plotted the trend in a 1999 white paper, says light output per LED package rises about 20× per decade while cost per lumen falls about 10× per decade. That's roughly a doubling of lumens every 28 months and a halving of cost per lumen about every three years. The trend comes from several improvements multiplying together: converting electrons to photons more efficiently, getting more light out of the chip, handling more current, removing heat better, converting blue light to white more efficiently with phosphors, and better manufacturing yield. Because these factors multiply, gains compound, producing a straight line on a log plot. Like Moore's law, it's an observed trend, not a physical law, and it slows as each efficiency approaches a hard limit—for example, conversion efficiency can't exceed 100%.

Haitz's Law is an empirical regularity, plotted by Roland Haitz in a 1999 OIDA white paper on semiconductor lighting: lumens per LED package grow by about 20× per decade and cost per lumen falls by about 10× per decade (≈28-month doubling of output, ≈3-year halving of cost). The trajectory reflects simultaneous gains along multiplicative loss-reduction axes: internal quantum efficiency, light-extraction efficiency, current capacity per device, package thermal management, phosphor conversion efficiency for white LEDs, and manufacturing yield. Because the overall figure is a product of these terms, improvements compound into a log-linear trend, which held well from the late 1960s to the mid-2010s. Like Moore's law, it is a product-class doubling-time regularity from stacked engineering progress rather than one physical mechanism. It is used as a roadmap baseline: an application priced out by a factor N becomes affordable in about 3·log₂(N) years; below-trend progress flags a binding constraint worth research investment; and it guides capacity planning. The trend must decelerate as terms approach ceilings—internal quantum efficiency cannot exceed unity, and phosphor conversion caps white LEDs near 80% wall-plug efficiency, versus about 30% for commercial packages in the mid-2010s—so it is time-bounded, not permanent.

Scope of Application¶

Haitz's law applies within optoelectronics, to device classes whose performance is the multiplicative product of several independently-improving loss terms under sustained R&D.

LED device design — the home turf, where roadmaps scope new chip generations.
Solid-state-lighting industry strategy — capacity and fabrication-line planning on the doubling schedule.
Solid-state-lighting public policy — efficiency mandates whose practicality turned on LED cost crossovers.
Optoelectronic cousins — infrared LEDs, UV-LEDs, and laser diodes at different rates and offsets.
Display economics — micro-LED and mini-LED backplanes inheriting a slower version of the trajectory.

Clarity¶

Stating Haitz's law turns a vague sense that "LEDs keep getting brighter and cheaper" into a quantitative baseline against which any chip generation can be judged. That makes progress diagnostic: a below-trend generation is evidence of a binding constraint deserving research, not just disappointment. It makes the adoption horizon a calculation rather than a guess, and — framed as a law named for an observed trajectory — it keeps sharp that this is a time-bounded regularity, not a permanent physical guarantee.

Manages Complexity¶

An LED package's performance is the joint product of half a dozen loss terms evolving on their own timescales. Modeling that coupled system would mean tracking all six. Haitz's law collapses it onto a single scalar: because the terms enter multiplicatively, their compounded effect is a log-linear trajectory with a characteristic rate constant on each axis — lumens-per-package doubling every ~28 months, cost per lumen halving every three years. The engineer carries those rate constants and a baseline, reading adoption timing, the below-trend diagnostic, and the eventual deceleration off one line.

Abstract Reasoning¶

The law licenses crossover-date arithmetic (computing when an application priced out by factor N becomes affordable), diagnostic reasoning (a below-trend generation points at the binding loss term), boundary-drawing (refusing to extrapolate the line past the thermodynamic ceilings), and a compounding-from-multiplicative-axes account explaining why the line held across decades and where it becomes vulnerable when several terms saturate at once.

Knowledge Transfer¶

Within optoelectronics Haitz's law transfers as mechanism — the log-linear line, the crossover arithmetic, the below-trend diagnostic, and the ceiling-deceleration boundary carry to IR-LEDs, UV-LEDs, and laser diodes, each a genuine instance. Beyond it, the famous companions (Moore's, Swanson's, Wright's laws) are not exports but siblings under a shared parent — the experience/learning_curve prime — differing only in product class and slope. The cross-domain lesson carries that parent, including its eventual deceleration; the LED-specific loss axes and figures are optoelectronics furniture that stays home.

Relationships to Other Abstractions¶

Current abstraction Haitz's Law Domain-specific

Parents (1) — outgoing typed relations

Haitz's Law is a kind of Learning Curve Effects Prime

Haitz's Law is a Learning Curve Effect specialized to LED packages whose light output rises and unit cost falls predictably with cumulative engineering and production experience.

Hierarchy paths (3) — routes to 3 parentless roots

Haitz's Law → is a kind of Learning Curve Effects → is a kind of Increasing Returns

Show alternative paths (2)

Neighborhood in Abstraction Space¶

Haitz's Law sits in a sparse region of the domain-specific corpus (86^th percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.

Family — Unclustered & Miscellaneous (2769 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-10-25