Catching up with the trend of AI development, light chip startups have raised hundreds of millions of dollars in investment capital
A few years ago, computer chips that used light for processing were only available in research projects, but now these projects are bearing fruit, and startups addressing the challenges of this technology are emerging. raised millions of dollars in investment.
Most recently, Ayar Labs, a startup developing technology called silicon photonics (silicon chips), said last Tuesday it had raised $130 million from investors including the tech giant. Nvidia chip giant.
Ayar Labs’ light emitter is called SuperNova.
But they are not the only ones. Data from PitchBook shows that last year, light silicon chip startups raised more than $750 million, double that of 2020. Meanwhile, in 2016, this figure was only $18 million.
After decades of exponentially increasing power, semiconductor silicon chips are slowly reaching their physical limits, when the size of each transistor is now only the radius of a few atoms.
Meanwhile, the demand for high-performance computing is increasing with the development of AI and big data processing. Ayar Labs CEO Charles Wuischpard said:AI is growing at a crazy rate and taking up an increasingly large share of data centers. The challenge of data migration and energy consumption in data migration is a very, very big one.”
This challenge lies in the fact that large machine learning algorithms can use hundreds or thousands of chips at the same time to compute and create bottlenecks in the data transfer rate between chips or servers when using the line. electricity to process calculations.
Ayar Labs optical chiller called TerePHY
For decades now, light has been used to transmit data over fiber optic lines, including undersea cables. However, shrinking them to fit in chips is a bigger challenge than shrinking transistors. Even so, research on them is opening up enormous potential.
PitchBook Emerging Technologies Analyst Brendan Burke expects light silicon chips to become ubiquitous hardware for data centers by 2025 and is expected to reach $3 billion in size. equivalent to the size of the AI graphics chip market by 2020.
Not only connecting semiconductor chips, startups using light silicon chips in quantum computers, supercomputers, and chips for self-driving cars have also raised large funds.
So far, the startup PsiQuantum has raised $665 million, although the promise of quantum computers using light chips is still many years away from becoming a reality.
Lightmatter, a startup that develops processors that use light to accelerate AI workloads in data centers, has also raised $113 million and will launch its chip later this year.
Luminous Computing, a startup developing supercomputers using light silicon chips, backed by Bill Gates has also raised $115 million.
The PsiQuantum silicon dish, jointly produced by PsiQuantum and GlobalFoundries, can hold up to thousands of quantum devices, including single light detectors.
Not only startups, even semiconductor manufacturers are starting to keep an eye on the inclusion of light in their semiconductor chip manufacturing technology. GlobalFoundries said it is also working with PsiQuantum, Ayar and Lightmatter to help develop a light silicon chip platform that other partners can use. The platform launched this past March.
Refer to Reuters
at Blogtuan.info – Source: genk.vn – Read the original article here
