Explaining the Chip Industry Boom

What caused the advent of NVIDIA’s reign? Why did A.I have a growth spurt? Why are we where we are? Analysing the history of growths and declines gives us valuable lessons on what to expect in the future. We’ll take you on a quick journey through the 20th century to the 21st, mapping the trajectory of progress from the first transistor, to the release of GPT-4 today.

1947: The point contact transistor, invented by John Bardeen and Walter Brattain was the first true transistor ever created. In June 1948, they continued on to invent the junction-type transistor, used mostly for audio in pocket radios, though the point-contact would remain the fastest available until the invention of the surface barrier transistor in 1953.

1956: Ivy League research university Dartmouth College in New Hampshire officially establishes Artificial Intelligence as a research discipline.

1964: Very-large-scale integration (VLSI) technology is born with the MOS integrated circuit, housing more than 10,000 transistors within one chip.

1970s & onward: VLSI grows exponentially, from tens of thousands, to today’s billions. Integrated device manufacturers (IDMs) like Intel and Samsung stood as high-cost industry standards. Companies such as Apple, AMD and NVIDIA started smaller by adopting the fabless model, selling their designs without their own fabrication facility. Intel remains one of the few chip designers with its own production facilities.

1980s: The NEC μPD7220 becomes the first PC with a dedicated large-scale integration (LSI) graphics processor. The term GPU wouldn’t be coined until 1994 by Sony & its 32-bit Sony GPU, with NVIDIA popularising it by declaring its GeForce 256 the world’s first GPU.

2009: Satoshi Nakamoto creates Bitcoin, sparking a growth in computing demand with its Proof-of-Work model. Mining demand gradually switches from CPUs to GPUs.

2012: A research team from the University of Toronto uses deep learning to achieve a computer vision error rate of under 25% for the first time during the ImageNet challenge. The alumni of ImageNet go on to become leaders in tech.

2013: Mining hits its first major tech ceiling. ASICs are invented with the sole purpose of mining Bitcoin, and remain the leading tech in mining until today.

2016: DeepMind’s AlphaGo A.I defeats the human world Go champion, Lee Sedol, putting A.I solidly in the public perception.

2017: OpenAI is founded. Venture capital funding in the U.S. reaches its peak, and will stay as the highest in the world until 2021.

2018: A.I research explodes. Europe publishes the largest number of A.I papers in that year.

2020: GPT-3 is released to the public, solidifying the A.I boom. That same year, PC gaming records were being broken at the same time Bitcoin prices hit new peaks, all while GPU manufacturing took a hit from COVID. NVIDIA’s RTX 3080 sells out, followed by AMD’s Radeon 6000 just two months later. This shortage would carry on into 2021.

2021: OpenAI announces DALL-E.

2022: MidJourney is born, followed by Stable Diffusion.

2023: Tech giants like Google & Meta race to meet its A.I competitors head on with Gemini and Meta AI/Llama. NVIDIA sells out of its A.I chips & H100s and announces that there won’t be more until 2024. NVIDIA has a stranglehold on A.I chip production due to its excellent CUDA ecosystem for software programming, making it a costly choice for those reliant on it to switch to other computing companies.

2024: Sora previews its text-to-video results to the world. GPT-4 is released.

2025: Multimodal A.I like Gemini 2.5 are now the new baseline, where users can turn text inputs into videos or find answers from a photograph.

Judging from the trajectory of A.I’s progress alongside the skyrocketing of NVIDIA’s, it’s more than likely that the chip industry boom still has space to expand. As of today, transistor technology is slated for retirement: developers are already aiming to surpass the many limitations of today’s computing, with 3D manufacturing being only one of those potential directions.

The movements of history dictate the progress of the future: AI won’t be done with its explosive growth anytime soon.