“AI Hardware Enters its ‘Perform or Go Home’ Epoch with a Chuckle!”
“AI Hardware Is in Its ‘Put Up or Shut Up’ Era”
“It’s been a decade since AI started its current boom cycle. In that time, startups in the space have seen nearly $40 billion in funding. A lot of that cash, however, has gone towards solving the compute and data problems associated with machine learning—the heart of AI.”
Well, well, well. It’s been a full ten great years since AI embarked on its trendy growth spark. And during that span, the lucky startups in this arena basked under the golden spotlight of nearly $40 billion in funding support. Now, wouldn’t some business sectors give an arm and a leg for a sliver of that pie! But, golly gee, here’s a shocker- an enormous slice of that tremendous funding cheese just got funnelled into ironing out computing and data glitches associated with machine learning, the pulsating heart of AI. How surprising!
“Despite all that investment, however, most of the devices the AI industry relies on—the GPUs, the TPUs, the FPGAs—are as old as the beats of rock ‘n’ roll.”
Despite that bountiful bounty of investments, it seems the AI sector’s reliables – GPUs, TPUs, FPGAs – are archaic relics, as dated as the tune of rock ‘n’ roll. Indeed, cue the grand entrance of ironies. With all that hefty funding, you would expect them to at least have spangly new toys!
“In retrospect, it seems inevitable that the AI industry would have a reckoning with its hardware. Machine learning has never been a very efficient process.”
Looking back, it’s clear as a bell that a standoff between the AI sector and its hardware was only a matter of time. By no means has machine learning been the poster child for efficiency. Did we really expect a heated or rather, cooled, love affair between the two?
“So, to compete in the future, companies must make chips that are AI-first or face an eternity on the sidelines.”
Thus, the gauntlet has been thrown – it’s time for companies to forge AI-centred chips or embrace an eternity in the dreaded margins. A chilling prospect indeed, no company wants a ride on that train.
“In 2018, when AI hardware was in its infancy, the industry saw more than 100 new chips. You see, things are moving fast.”
In the days of yesteryear – 2018, to be exact – when AI hardware was still toddling about, the industry beheld an influx of over 100 spanking new chips. Yep, the gears in this sector spin at a blink-and-you’ll-miss-it speed.
“If the AI and machine learning sectors have their ducks lined up properly and keep evolving at this accelerating pace, we might just encounter AI that is virtually omnipresent and unfathomably faster by CES 2025.”
If the stars align for the AI and machine learning clusters, maintaining this meteoric pace of progression, who knows? We might just stumble upon a nearly ubiquitous, mind-bogglingly swift AI at CES 2025. Science fiction, is that you?