Mark Kennis from Stocks Down Under and Pitt Street Research says that if investors want to keep close tabs on the sector, they should keep an eye on the Philadelphia Semiconductor Index.
“It’s by far the most important index for semiconductor stocks,” he told the Stocks Down Under Semiconductor Conference last week.
“It’s a basket for US based stocks listed in the US, equipment players, fabs companies, and a few foundries.
“Right now, it’s near all-time highs, and it’s a really good index to watch because typically the investors in this space look about six to nine, sometimes even 12 months ahead.
“So, if you see a turning point in stocks, that’s usually a good indicator that the industry is turning, maybe going back to an oversupply situation again.”
Here are how the stocks presenting at the conference see 2022 playing out.BrainChip (ASX:BRN)The company is by far the biggest player by market cap in the sector, with its Akida 1000 Neuromorphic System-on-Chip able to learn in real time, mimicking neurobiological architectures and working like a mini brain.
The company made a swathe of announcements in November, entering into a
four-year licence agreement for its Akida IP with MegaChips – a pioneer in the Application Specific Integrated Circuit (ASIC) industry.
BrainChip also completed
functionality and performance testing of Akida, which achieved performance and lower power consumption results.
Company founder and chief technology officer Peter van der Made –
who handed over the CEO reins to Sean Hehir last month – told the conference that the extremely low power consumption is good for the planet compared to current AI computing which generates huge amounts of greenhouse gases.
“If you look at the collected data centres of the world, they emit around 600 million pounds of greenhouse gases,” he said.
“An Akida 1000 is estimated to be around 97-99% more energy efficient processing.
By 2030 if we continue the way we’re going, data centres will account for 30% of global electricity – and Akida can solve that problem with its low processing power.
And Akida is independent from the cloud, doesn’t require an internet connection to operate and can learn in milliseconds as opposed to weeks with traditional AI.
“With Akida, it’s completely independent of the cloud, all the information is processed on the chip itself, so it’s very secure, whatever stays within the chip itself cannot be hacked,” van der Made said.
The production chips are now being integrated into complete PCIe and Mini-PCIe boards, with these boards and evaluation systems being shipped to Early Access Customers for further testing and verification to see how they work as part of customers’ products.
(20min delay)