Semiconductor technology has reshaped the history of music consumption

From the 1877 invention of the phonograph to the rise of generative AI, semiconductor advancements have transformed music from momentary sound into digital data. SK hynix is now supporting the AI music era through High Bandwidth Memory and plans to begin mass production of 321-layer 2-terabit QLC NAND flash in August 2025.

Music, which has accompanied human history, has changed its form according to the era. From the age of oral tradition, where sound disappeared once a song ended, to the current era of AI music that creates new tracks with a single prompt, technological progress has redefined the very meaning of music. A story shared by SK hynix details how semiconductor technology has permeated the history of music, from the recording of sound to the process of generative AI creating music.

Recording and Amplifying Sound: Music Enters the Living Room

In an era before technology existed to physically record sound, music was passed from mouth to mouth or preserved only in sheet music. While sheet music captured melody and rhythm, the sound itself could not be heard again once a performance ended. Change began in 1877 when Thomas Edison invented the phonograph, a mechanical device that engraves sound vibrations into grooves on a cylinder. Later, in 1887, when Emile Berliner invented the gramophone, which engraves sound onto discs, music such as classical and opera could spread beyond concert halls and into homes.

However, early mechanical devices were limited by volume. Before electrical amplification technology existed, the sound from horn speakers struggled to fill even a single room. This was resolved by the vacuum tube, which emerged in the early 20th century. In 1906, Lee de Forest invented the triode vacuum tube, opening the way to amplify minute electrical signals, and with the development of amplifiers and speakers, music became louder and clearer. When commercial radio broadcasting began in the 1920s, music spread into living rooms through radio waves.

While vacuum tubes had disadvantages such as high heat generation and fragility, the listening environment for music underwent another upheaval in 1947 when researchers at Bell Labs invented the transistor using semiconductors. Transistors were smaller and operated at lower power than vacuum tubes, leading to the miniaturization of electronic devices. Thanks to the 'TR-1', the world's first commercial transistor radio released in 1954, the younger generation could enjoy music outside the home, expanding music from the living room into the realm of mobility.

Digital Transformation and the Era of 'Ownership' Opened by Flash Memory

The emergence of the integrated circuit (IC), a core semiconductor technology, shifted the way music was consumed toward the concept of "ownership." The integrated circuits implemented by Jack Kilby and Robert Noyce in the late 1950s gathered complex components onto a single small chip, allowing for sophisticated electronic devices. Later, with the appearance of portable cassette players like the Walkman, people began recording their favorite songs onto tapes to carry with them. However, analog media like tapes had the limitation of degrading sound quality during repeated playback.

This limitation was overcome by Pulse Code Modulation (PCM), which records sound waveforms as 0s and 1s. In 1982, when Sony and Philips commercialized CDs using the PCM method, music finally became digital data. In particular, the innovation of the Korean semiconductor scientist Dr. Dawon Kahng served as a decisive catalyst in accelerating the digital music era. The MOSFET announced by Dr. Kahng in 1960 became a fundamental building block of modern semiconductors, and the floating gate structure announced in 1967 became the foundation for flash memory, which retains data even when power is turned off.

With the addition of MP3 compression technology in the 1990s on top of this technical foundation, an era opened where thousands of songs could be contained on a single palm-sized chip. Subsequently, the popularization of smartphones and wireless networks in the 2010s moved music beyond individual ownership into the streaming era, where music is retrieved in real-time from music stored on data center servers. Now, music exists not in individual devices, but within massive digital libraries concentrated with NAND flash-based SSDs for data centers and CPUs and GPUs for servers.

The Engine of the AI Music Era: HBM and Ultra-High Capacity Memory

The music industry is currently facing another massive turning point. The era of generative AI, which creates new music simply by inputting a prompt, has arrived. The Transformer model, which appeared in 2017, learned complex relationships between data to master patterns of sound and music, extending into the realm of composition. To drive these massive AI models, computational power capable of rapidly processing vast amounts of data is essential.

High Bandwidth Memory (HBM) plays a key role in this process. SK hynix, which developed the world's first HBM in 2013, supports AI music generation by vertically stacking multiple DRAMs to supply data to GPUs without interruption. Additionally, SK hynix is maximizing data center storage capacity by planning to complete the development and begin mass production of 321-layer 2-terabit QLC NAND flash, which stores 4 bits per cell, in August 2025. From April 2026, the company plans to expand its application range by starting to supply PC SSDs equipped with this technology.

While past semiconductor technology focused on recording and moving music, evolved memory technology now functions as a new instrument that imagines and creates music. Music, which once existed as momentary sound, has now become data, serving as the driving force to realize human imagination into reality.

Source: original video (YouTube)