U.S. edge chip designer Blaize appointed Cisco alum Stephen Patak as its first chief revenue officer (CRO).

Blaize debuted on the Nasdaq via a SPAC merger last January. Widely reported as the first AI chip startup to ever go public, Blaize has since signed deals with the likes of Nokia’s Singapore arm to power practical and physical AI systems for distributed environments across Asia-Pacific (APAC).

Continuing in this spirit, the firm said Patak would lead its global revenue and go-to-market execution as it looks to expands AI deployments across public and private sector markets.

The exec joins Blaize from a two-year CRO stint at Nile. Prior to joining the cloud-native networking vendor, Patak served three years as CRO for smart city tech player Ubicquia, while previous roles include VP and SVP roles for location platform Here Technologies.

His networking sales experience includes Extreme Networks, where he led as VP and then SVP for the Americas region. Perhaps most pertinently to Blaize, Patak cut his teeth at Cisco for more than 14 years, including five years in the Asia-Pacific, Japan, and Greater China region, working from Singapore.

On his appointment, Patak said: “Blaize is working with cloud service providers, next-generation AI cloud operators, and infrastructure partners across both public and private sectors, to make AI part of core platforms, not just an experiment. That’s a compelling stage to join, and I’m looking forward to working with the team as Blaize continues to expand.”

In an interview with SDxCentral, Blaize CEO and co-founder Dinakar Munagala explained the California-based firm has a globally distributed team and customer base, with core corporate functions including sales based in Sacramento and San Jose, with field sales teams based in the United Arab Emirates (UAE) and Saudi Arabia. A data science team can be found in North Carolina, while a sizable India development center handles core processor architecture, drivers, and compilers.

Blaize’s system on chip (SoC) and board design teams are located in the U.K., Kings Langley and Leeds to be precise, with talent in the former mainly sourced from MIPS Technologies after a period of consolidation between MIPS’ U.S. and Chinese owners. The company currently stands at 230-strong worldwide.

“As far as the sales funnel and the customer traction goes, it's quite global,” Munagala explained, referencing deals such as a $120 million smart city-focused venture targeting AI infrastructure in East and Southeast Asia.

“We publicly announced contracts in India with Yotta,” he added. “Of course, there’s the Nokia plans, with our initial focus on Asia. There’s a large customer funnel in the Middle East, then North and South America.”

Joseph Sulistyo, SVP of global marketing, explained that besides North America, the threadline between those target territories is simple: cost is still critical, especially around power efficiency.

“They're very cost-sensitive. If it's a few cents, a few pennies, basically, that will mean a lot, right?” the SVP said, adding U.S chip export restrictions to China and Israel also have their part to play.

“This is where we find that our chip actually can provide those alternatives, and we can actually complement the existing chip," Sulistyo said. "What I’ve learned from some of the biggest enterprises out there, they always want to be fault-tolerant, and so they never want to go with just one chip. Those reasons combined provide the ability for us to be able to serve this underserved market."

‘Trailblaizing’ ahead

Blaize's edge computing platform centers around its 1600 SoC AI processor, which is built on graph streaming processor (GSP)-based hardware engineered to drive inference workloads within AI clusters, enabling the processing of localized data such as sensor streams at the edge.

With flagship chip the Pathfinder X1600E able to run AI inference models at lower power consumption than traditional GPU-based systems, Munagala explained Blaize’s USP for the burgeoning physical AI market, which Nokia aims to capitalize on with its help.

“Most of the AI compute today is happening on massive training clusters and throughput computing. So latency is not as important; you can do things offline, batch training, etc., and that's where GPUs are very good,” the CEO said.

“But if you look at industries like automotive and so on, they need low latency in real time. A smart car needs to make a decision fast, for example," Munagala continued. "In our technology, the GSP is very suitable for low-latency real-time applications, like how a camera can infer a threat and take some action, or a drone, or robotics used in factory automation.”

As such, the automotive industry drove Blaize’s initial growth after it was founded in 2011 by Munagala alongside Ke Yin and Satyaki Koneru, who together spent a combined three decades in Intel’s graphics division. Initially known as Thinci, Blaize has raised more than $330 million in funding to date, with automotive makers as its primary investors.

With Patak as its sales chief, Blaize’s next steps include operationalizing AI for clients as well as bringing down cost of implementation.

“Our goal is to make it almost as you would buy a PC or a laptop, bring it down to that level of expense that a business should be able to afford AI and make use of it,” Munagala said.

There may also be room to expand its Nokia partnership beyond Singapore, but first, Blaize wants to establish the APAC deal as a “beachhead,” according to Sulistyo, with Nokia’s scale and size a boon for the growing firm.

Sulistyo explained that the two companies had common partners, which ultimately led to the Nokia Solutions and Networks Singapore deal.

That arrangement proves to Nokia watchers that the telecom vendor is not beholden simply to Nvidia chips since last year’s landmark deal with the company. More important than that deal, in Sulistyo’s view, is Nokia’s changing status in the current era, moving away from telecom and into the true AI space from its previous position on the parameters.

“With Nokia, we talked about a hybrid AI platform, with not only our own chip, but also a GPU like Nvidia and complementing those together as an integrated system," Sulistyo said. "And we can also build an application on top of it, and we can actually also extend API services, much like a cloud service provider. They were intrigued, and said, ‘You know what, why don't we try something?’

“So that's actually how this collaboration started, kind of building into what their vision, what their roadmap would want to be, and also to have kind of a more added value than just as a networking provider or a part of telco ecosystems," Sulistyo continued. "Basically moving really into the neocloud space, and the [colocation] data center providers are paying attention to it.

“There's going to be more things that we're going to unveil in the upcoming months," the SVP promised.