Blaize raises $71m to take on Nvidia from Europe
US AI chip and tool developer Blaize has raised $71m to take on the automotive market, bringing the total raised to $155m.
The Series D funding from investors including automotive tier one supplier Denso will see development of AI chips as the central controller. This will be led by the UK design teams in Kings Langley and Leeds, Dinakar Munagala, CEO of Blaize told eeNews Europe.
“This funding will drive our move into EVs in automotive,” he said. “We have been working with the automotive industry to build an automotive grade product,” he said, pointing out that the design flow for the chips and compiler was compliant with the ISO26262 safety standard from the start.
The first step is consolidating some of the 150 microcontrollers in an electric vehicle into a smaller number of domain controllers, typically four or five. This would include a dedicated AI chip from Blaize for battery pack monitoring and optimisation and powertrain control, for example.
“We have been getting requests for using cell level optimisation of the battery- that’s a huge thing – there’s different parts of the problem and that’s an interesting area,” he said. “One OEM is looking at the battery, another is looking to combine this with other functions in a domain controller.”
“In the longer term our aim is central compute in 2029 and we have engaged with customers for this time horizon,” he said. “A central processor is where all our OEM customers are heading and given the validation from the auto industry with Denso we are committed to delivering value to our customers in that space. That fundamental technology is all about efficiency, we are about five times more efficient than our competition,” he said. That competition is currently Nvidia with its Drive system, although other chip developers are also aiming at the same central AI controller market.
Related Blaize articles on eeNews Europe
- Code-free tool for edge AI launched
- AI chip startup taps two UK design teams
- Startup details architecture of GSP edge AI chip
- AI boards launched in module and PCIe formats
- The year edge AI took off
“Blaize System on Chip for automotive edge and central compute functions are accelerating electric vehicles and future architectural ambitions of automotive OEMs,” said Tony Cannestra, Director of Corporate Ventures at Denso. “With substantial power advantages making EVs more efficient and economical, Blaize SoCs offer best in class performance with lower power across in-cabin, out of vehicle, and autonomous operations, enabling a streamlined architectural evolution to centralize compute.”
“The automotive design team in the UK we acquired in Kings Langley and Leeds are doing the system level design and driving the applications for AEC [automotive qualification] and safety critical software and the ASIL [safety] roadmap, especially around the design practices,” said Munagala.
Next: Edge IoT roadmap with AI Studio
The funding will also be used for Blaize to develop the AI studio software for edge IoT applications and develop vertical marketplaces to download a wide range of AI applications.
“We have an exciting roadmap, and the opportunity that we have is for transition from AI 1.0 to AI 2.0,” he said. “is retrofit what you have to CPUs ad GPUs, which is really hard to do, and needs data scientists and ML engineers. 2.0 is all about ease of adoption. AI studio is an environment where it’s really easy to build AI application in a code free manner and the third element is allowing anyone to build AI, doctors, bankers. It’s a DOS to Windows kind of a leap.”
The funding will also be used to add another 150 people in the next 18 months, he says.
The $71m Series D will take the company towards break even and will be used to pay for future manufacturing capacity
“Our 2021 and 2022 revenue comes from product today, and there is a server grade part in 18 months and that will start to ship in 2023 and if there is future investment required we will go back to investors,” said Harminder Sehmi, chief financial officer. “This $71m allows us to scale automotive and AI studio road map based on the current chip.”
The company works with both TSMC and Samsung as foundries, with the first chip on Samsung’s 14nm industrial process, rather than commercial grade. This has automotive option which allows devices to be tested in the field in vehicle trials at the moment.
The company has pre-purchased capacity in 18 months time. “For 7nm process you are looking 18 to 24 months lead times for the packaging substrate,” said Sehmi.
The company also brought production of cards and modules back from Asia ahead of the pandemic. “We were alerted to the potential of something nasty around the corner by [Singapore investor] Temasek and we started to take action, bringing contract manufacturing of the cards back to the US, which added slightly to the cost, and we keep putting orders into the factory to reserve capacity,” said Sehmi. “Some parts have a two year timeline so we pre-purchase that, and with contract manufacturing we are pre-purchasing certain long lead time items over the next two years.
Temasek was also part of this investment round with Denso alongside fund manager Franklin Templeton as a new investor.
“We are pleased with the participation of new and existing premier investors in the Series D round,” said Munagala. “We are well positioned to accelerate our next generation products and sustain our lead in delivering integrated hardware and software offerings that enable true value creation for our customers.”
Other articles on eeNews Europe
- Ascenium raises $16m for next generation software defined processor
- Magna buys Swedish sensor maker Veoneer in $3.8bn deal
- Astonishment at failing UK battery strategy
- Oxford PV and Meyer Burger falling out threatens Europe’s solar industry
- ARM shows first plastic M0+ microcontroller
- AI tool cuts 3nm chip design times
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
