It's one thing for a large corporation to partner another in order to build a total solution for a product, but doing the same with a startup seems somewhat harder.
This is where leadership can play a role. The instinct of finance and legal teams as well as business units and others will be to point to the weaknesses inherent in any small company – lack of capital, team size and execution risks – but if the leadership at the corporation is strong enough it can identify the potential opportunities and benefits of working with best-in-class collaborators on a big problem.
Japan-headquartered materials group TDK has made just such a decision with an investment by its corporate venturing unit, TDK Ventures, in US-based machine learning chip developer Groq.
TDK has numerous partnerships with other large corporations, such as Qualcomm for the thinking and communication parts of robots, or Nvidia and Google in other specific areas. Groq, however, offers TDK the chance to define the roadmap together from the top down, in a strategic opportunity for artificial intelligence (AI) in areas such as autonomous driving.
Groq’s chip, or tensor streaming processor (TSP), can perform 1 quadrillion operations per second, which is a large number but basically means it is very good at neural network training and inferencing, which is helpful for areas such as self-driving cars where the volume of input is predictable, according to TechSpot.
The company came out of stealth in November and has backing from venture firm Social Capital and now CVCs. TDK’s high-performance passive components and power management solutions appealed to Groq as it tries to scale up its manufacturing.
Jonathan Ross, Groq’s CEO, used to help develop tensors at Google and said: “Through its investment and support, TDK is helping Groq bring these capabilities to customers concerned about ML [machine learning] compute power demands and energy consumption.”
And the energy consumption focus is an additional advantage to TDK Ventures as the power used by Groq’s TSP is an order of magnitude lower than for peers. Nicolas Sauvage, managing director at TDK Ventures, said: “Groq brings the highest AI compute for a given amount of energy, typically measured as tera operations per second per watt, demonstrated by its first product, the TSP100, with further projected improvement in their next gen[eration] platform.”
Data centres use an estimated 200 terawatt hours (TWh) each year, or about 1% of global electricity demand, contributing 0.3% to overall carbon emissions, though they could reach a double-digit percentage of the world’s electricity by 2025. If Groq is successful it could bring power use back down to single figures again, Sauvage said.
This is impact on a global stage.
