Tech Brew keeps business leaders up-to-date on the latest innovations, automation advances, policy shifts, and more, so they can make informed decisions about tech.

Tech Brew

Semiconductors

Jordyn Grzelewski

Tricia Crimmins

Patrick Kulp

techbrew.com

You’re tired of reading about tariffs. Honestly, we’re tired of writing about them. But amid the ever-shifting landscape of 

, and Truth Social missives, where do things stand 

, the Tech Brew team homed in on the ABCs of tech industries—automobiles, batteries, and chips—to illustrate how tariffs are impacting those sectors, the US economy, and your wallet.

? According to Art Wheaton, director of labor studies at Cornell University’s ILR School, the situation is “still as messy now as we were when we began.”

“There’s not a lot of clarity,” he told us late last month. “There is no final deal for Mexico. No final deal for Canada. There’s no final deal for China. So everything is still in flux for three of the biggest trading partners we have.”

In April, the Trump administration implemented tariffs of 25% on imported vehicles. Since then, the US has announced trade deals to lower auto tariff rates for several trading partners, including the UK, Japan, South Korea, and the EU.

“It’s really created a lot of confusion—which, if you ask the current administration, they like,” Wheaton said. “They like keeping everybody on their toes, but the auto industry hates uncertainty.”

In an August update on the auto tariffs landscape, Cox Automotive reported that “automakers have theoretically racked up more than $25 billion in tariff obligations so far,” based on “the average imported vehicle being charged roughly $5,200 at the US border.”

Overall, automakers and dealers appear to be absorbing at least some tariff-related costs rather than passing them on to consumers in full, according to Cox.

“But that won’t last,” Erin Keating, executive analyst at Cox, 

. Cox analysts forecast a 4%–8% YoY increase in retail prices by the end of the year.

Meanwhile, auto manufacturers are feeling the effects on their bottom lines, with companies like General Motors, Ford, and Hyundai reporting tariff-related hits to profitability in Q2. The biggest US automakers reported that tariffs hurt their collective profitability to the tune of $11.7 billion in the second quarter, Yahoo Finance 

“The auto industry is clearly under pressure,” Keating wrote. “Tariffs are increasing costs across the board for automakers, dealers, and consumers—not only for new vehicles, but for aftermarket parts and insurance costs.”

Cox analysts expect the average transaction price of a new vehicle in the US to surpass $50,000 this year, and for new-vehicle sales to be lower than in 2024.

And affordability is already a challenge in the auto industry: Yahoo Finance cited Atlanta Fed data indicating that car prices are growing six to eight times faster than household incomes.

Experts have noted that some other recent federal policy changes, like the rollback of vehicle fuel efficiency standards, could be tailwinds for domestic automakers.

“GM, Ford, Stellantis, and others can now make as many pickup trucks and large SUVs as they want,” Wheaton said. “So they’re able to absorb some of those tariffs because they think their profits are going to do better with a different vehicle mix, much worse [mileage] per gallon but higher profit margin.”

In the last few months, it’s felt like tariff uncertainty has taken a backseat to 

 in the green tech industry. But for battery manufacturers—which were able to 

 their 45Y and 48E tax credits—the tariff war has hit hard, forcing them to make tough decisions about where they source their battery cells.

Rithy Chhean, VP of technology at BoxPower, told Tech Brew that “when the price war escalated, [tariffs] became all-consuming” for the company, which produces batteries in addition to solar powered microgrids. While portions of BoxPower’s hardware is domestically produced, the company—like many other battery manufacturers—uses battery cells from China.

“Over the course of three months, it went to almost a 150% tariff,” Chhean said. “We didn’t have special terms in our contracts to say, ‘Here’s the tariff costs, and if the tariffs increase, you have to pay it.’ Some customers were able to deal with it. Some customers, you’re not able to do anything. Someone has to pay that difference.”

BoxPower’s forthcoming LFP battery, the ABX‑110‑BABA, fulfills Build America, Buy America (BABA) requirements, meaning at least 65% of the battery's materials were produced in the US. Even so, Chhean said BoxPower wasn’t able to find any cost-competitive domestic battery cells to use in the product.

“It’s cheaper to get cells from basically another country and pay the tariff costs,” he told us. “Even the companies that say they’re highly domestic, we find out that their cells or modules are also coming from China anyway, or somewhere like China.”

Though the Trump administration claims the goal is to increase “advanced domestic manufacturing capacity” through tariffs, doing so will take time in the battery industry. Critical mineral developers are working together to 

, and battery makers have plans to bring manufacturing plants to the US in the future, but right now companies like BoxPower are in a bind.

At the moment, Chhean said BoxPower uses a “hybrid approach.” The company still buys battery cells from China, but when possible, it buys from companies in other countries that have lower tariffs—even though the cost of the item might be higher. Above all, BoxPower says it’s working to be transparent about tariffs with customers.

“We’re trying to show what the tariff costs are,” Chhean said. “If the tariffs do increase, we will provide a very small markup, if any markup, on it because it’s not really fair to the customer, and we understand that.”

There’s plenty of uncertainty and confusion to go around, too, when it comes to Trump’s moves on 

, which power everything from consumer electronics to advanced AI chips.

 100% tariffs on semiconductors, a possibility his administration has been probing since April. But the announcement also contained a key exemption: Companies that have committed to investing in manufacturing in the US or are already doing so are not subject to the tariffs. The Trump administration has already 

 Apple as one of these companies. Some of the biggest players in the AI infrastructure boom—Nvidia, TSM—–may 

 about the details of these tariffs, which have yet to be formally rolled out. The Semiconductor Industry Association said last month that “we are eager to learn more.”

In such an uncertain environment, Lydia Clougherty Jones, a senior director analyst with Gartner, said companies will likely need to consider multi-country sourcing strategies and stay nimble with scenario planning and analytics.

“It is critical to align technology procurement decisions with the highest-priority disruption-proofing needs, including enabling executive teams to use real-time situational analysis to prepare for future short-term and long-term trade war disruptions,” Clougherty Jones said in an email.

 in running the chipmaking industry itself, most notably by having the US government 

 in Intel. Trump is also directly negotiating fees with Nvidia and AMD to continue selling chips in China, in what legal critics say 

Clougherty Jones said the US government’s more active role in setting industrial policy for the chip sector could ultimately speed AI development—a goal for both the government and the companies involved—but they may also end up at odds over differing motivations.

“Such public-private partnerships prioritize speed, resulting in the acceleration of AI R&D and infrastructure build out, including of data centers, without regulatory guardrails, as governments can selectively remove barriers to AI progress or deprioritize their enforcement,” she said.

“On the other hand, there can be a clash of priorities, as private companies are driven by profit motives and the goal of delivering maximum returns to their shareholders, whereas governments may prioritize advancement of sovereign-specific goals such as national security and strategic AI autonomy, leading to dissimilar drivers and measurements of what successful AI advancement looks like,” she said.

A truck passes by China Shipping containers at the Port of Los Angeles, California, on September 1, 2019.

How is tariff uncertainty impacting tech?

We focused on the automotive, battery, and chip industries to tell the tale.

How is tariff uncertainty impacting tech? 

The generative AI wave is still minting money for Nvidia, which has come to be seen as a bellwether for the longevity of hype around the budding technology.

Nvidia execs played up that role in an earnings call this week, touting the success of AI frontrunners—and Nvidia customers—like OpenAI, Google, and Meta in turning AI into a durable business. The company continued its 

 with a 262% YoY jump in revenue to $26 billion and a profit increase of more than 600% from the same period last year to $14.88 billion.

Nvidia’s ongoing success is fueled by companies of all kinds—but especially cloud giants—continuing to pour funds into building out an infrastructure to train and run massive foundation models. CEO Jensen Huang refers to these advanced data centers as “AI factories” of a “next industrial revolution.”

Nvidia CFO Colette Kress said cloud providers buying Nvidia chips are seeing an “immediate and strong return,” claiming that for every dollar they invest in hosting capabilities, they “have an opportunity to earn $5” in revenue over four years. And Microsoft and Google did indeed 

 this quarter, pointing to an ongoing business appetite for AI tools.

The momentum has continued even as Big Tech companies like Microsoft, Meta, and Amazon have also 

 their own AI chips to compete with Nvidia. Rivals like AMD and Intel are also 

In the call, Huang promised imminent dispatches of Nvidia’s 

recently announced Blackwell architecture

, with shipments starting next quarter and ramping up the following one. As to whether companies might be tempted to hold off on buying current offerings until the new, more powerful chip hits the market, Huang said the pace of the current AI race leaves no time for waiting.

“The next company who reaches the next major plateau gets to announce a groundbreaking AI. And the second one after that gets to announce something that’s, you know, 0.3% better,” Huang said. “And so the question is, do you want to be repeatedly the company delivering groundbreaking AI? Or the company delivering 0.3% better?”

Nvidia sought to show AI hype has legs in big earnings beat

The company continued its tear with another jump in revenue this quarter.

Kelcee Griffis

Congress’s massive investment in the domestic semiconductor industry is expected to pay off over the next decade.

 the Semiconductor Industry Association (SIA) and the Boston Consulting Group released Wednesday, which found that the US is poised to significantly grow its share of chip manufacturing both at home and abroad.

The country’s fabrication capacity will jump 203% from 2022 to 2032, SIA and BCG estimated. The country is also on track to expand its share of international manufacturing capacity from its current 10% to 14% by 2032.

The study notes that the projected growth is facilitated by the 2022 CHIPS and Science Act, 

 $39 billion in domestic manufacturing incentives and $11 billion for chip research and development. The legislation aimed to decrease American dependence on geopolitical rivals like China that supply crucial tech components.

“Effective policies, such as the CHIPS and Science Act, are spurring more investments in the US semiconductor industry. These investments will help America grow its share of global semiconductor production and innovation, furthering economic growth and technological competitiveness,” SIA board chair Rich Templeton said in a statement. “Continued and expanded government-industry collaboration will help ensure we build on this momentum and continue our next steps forward.”

 up to $285 million in funding through the act for the development of semiconductor digital twins, a technology that creates virtual replicas of physical objects.

US semiconductor manufacturing projected to skyrocket in wake of CHIPS Act

Study projects 203% fabrication capacity growth by 2032.

 have proven to be a popular tool for everything from 

. Now, the US government wants to use these virtual replicas of physical objects to supercharge chip manufacturing.

The Biden administration announced this week that it’s seeking proposals for projects that develop digital twins of semiconductors, in hopes of a more collaborative and efficient design process. These projects will be part of a new institute under the Department of Commerce, with up to $285 million in funding through the 

, Biden’s multibillion-dollar legislative boost to the domestic semiconductor industry.

The call comes as the generative AI craze has led to 

 toward improving this stage. The Biden administration said digital twins will help make this technology more viable, among other benefits such as increased remote collaboration and better testing.

“Digital twin technology can accelerate the costly and time-consuming work to develop the next generation of robust manufacturing for this extraordinarily complicated product,” Arati Prabhakar, director of the White House Office of Science and Technology Policy, said in a statement.

The CHIPS Act, passed in 2022, set aside $39 billion in domestic manufacturing incentives plus another $11 billion to be doled out for chip research and development. Funding for Commerce’s new Manufacturing USA Semiconductor Institute comes from the latter earmark.

 is to spur semiconductor innovation at home and limit dependence on geopolitical rivals like China, as a matter of national security.

Commerce’s announcement said funded activities will include “basic and applied research related to semiconductor digital twin development; establishing and supporting shared physical and digital facilities; industry-relevant demonstration projects; and digital twin-related workforce training.” The program will host a daylong meeting for potential applicants on May 16.

Futuristic microchip processor with lights on the blue background

Biden admin announces plan to shell out $285m for digital twins of semiconductors

The money comes from the CHIPS Act’s $11 billion R&D fund.

Justin Timberlake is slated to appear at San José’s SAP Center later this spring, but this week, a crowd of thousands filed into the home arena of the city’s NHL team to see a middle-aged man in a leather jacket talk about semiconductors.

That’s where Nvidia CEO Jensen Huang announced his company’s much-anticipated new generation of AI chips on Monday, aiming to prolong a hot streak for 

The company claims its new line of Blackwell graphics processing units (GPUs) can outperform its current top offerings, the Hopper architecture, manyfold across several metrics. The chip giant is betting that the new GPUs will extend its wide lead over competitors as the go-to source for processors that can handle giant AI models, a dominance that’s rocketed Nvidia to a 

“Ladies and gentlemen, I’d like to introduce you to a very, very big GPU,” Huang told the crowd of developers.

Despite all the fanfare, investors greeted the news with a lukewarm reception, with the share price dipping around 2% in morning trading on Tuesday. The slippage perhaps reflected just how sky-high 

 The keynote was part of Nvidia’s annual developer conference, running throughout this week, which some analysts are calling the “Woodstock festival of AI,” 

 to be described that way). Among the AI luminaries slated to make an appearance are executives from OpenAI, Google, and Meta, plus representatives of a host of well-known brands across sectors.

“One-hundred trillion dollars of the world’s industries is represented in this room today,” Huang said onstage.

Nvidia also sought to underscore its value to the tech industry, padding its new chip announcement with glowing quotes from an impressive who’s who of Silicon Valley bigwigs, including Mark Zuckerberg, Elon Musk, and Sam Altman. More than half of the names listed are reportedly 

 Nvidia’s lock on the AI chip market, but those bids are far from competitive yet.

 Huang said Nvidia’s new Blackwell chip can train AI with 2.5 times the performance of Hopper, and perform five times better at inference, or the actual running of an AI model once trained. The company also claims that Blackwell will require up to 25 times less cost and energy to run than Hopper.

Huang also touched on a host of other announcements throughout his two-hour keynote, from an upgraded system for autonomous-delivery vehicles and robotaxis to climate prediction tools. In flashy Silicon Valley fashion, he ended his speech with talk of a “new industrial revolution” accompanied by sci-fi-esque robots.

“A trillion dollars worth of installed data centers will become modernized over the next several years,” Huang said. “The computer of this revolution…[will be] Blackwell.”

Jensen Huang on stage at SAP Center during the keynote.

Nvidia announced its most powerful chip yet at ‘AI Woodstock’

The chip giant is looking to extend its dominance in the generative AI race.

Semiconductor technology was the most popular field for new tech inventions in 2023, according to new data analyzed by IP management platform Anaqua.

The firm crunched data related to nearly 350,000 patents the US Patent and Trademark Office granted last year, finding that virtual reality, 5G, and AI also led the way as topics for new tech developments.

Looking deeper into the development of AI applications, Anaqua found that most of the granted patents focused on “machine learning models, general neural network development, neural networks combination technology, neural networks for image and video recognition, and neural network training using backpropagation processes,” according to the firm.

IBM, Samsung Electronics, Alphabet, Microsoft, and Amazon ranked among the top companies that received AI-related patents, the report said.

“AI patenting continues to be strong, and now we’re seeing much more nuanced, sophisticated types of AI technology being patented in the US,” Shayne Phillips, director of analytics solutions for Anaqua’s AcclaimIP platform, said in a statement to Tech Brew. “The data shows new filings related to combining multiple neural networks together for even more powerful insights, and backpropagation network training, where AI is being used to train even newer AI models.”

Of the tech companies Anaqua tracked, Samsung scored the most new patents overall, securing more than 10,000 grants on technologies ranging from digital advertising to semiconductor manufacturing.

The other top patent grantees lagged quite a bit behind by volume: IBM secured about 4,000 new patents, while Qualcomm earned about 3,800 grants. Rounding out the top five innovators, Taiwan Semiconductor Manufacturing Company nailed down about 3,400 new patents, and LG Corp secured about 3,300 grants.

Semiconductors, VR, 5G, AI led new patent grants last year

IBM, Samsung among companies that secured most AI patents.

EdgeQ founder and CEO Vinay Ravuri wants to make standing up a mobile network as easy as plugging in a wi-fi router.

Whether it’s connecting an airport tarmac or a factory floor, most companies that need to provide internet access in commercial settings can’t simply turn on their own small-cell network. That’s partly because most microprocessors that power 5G networks require heavy customization to make them work in an enterprise context, Ravuri said.

At the same time, the main alternative—wi-fi networks—don’t usually offer the stability needed to support the high volume of traffic generated by large enterprises, according to Ravuri. That’s where EdgeQ, one of the few 5G chipmaking startups, hopes to step in.

“We’ve basically integrated what would be separate chips into a single chip. You can treat it as black box, and you can also make modifications if you have the wherewithal, but you don’t need to. It’ll work out of the box,” Ravuri said. “If I have a customer that wants a small cell…the ease of use is like wi-fi.”

Ravuri recently spoke with Tech Brew about how his startup plans to make private 5G networks more accessible by creating tech that’s purpose-built to support next-generation mobile infrastructure with minimal customization required. It’s a concept that’s generated buy-in since the company’s founding in 2018, raising more than $130 million from 16 investors, according to PitchBook.

“There is actually no good solution out there,” Ravuri said of companies that want to set up their own 5G network without repurposing chips designed for mobile phones or undertaking heavy-duty engineering to make their networks just…work.

The idea for EdgeQ bloomed during Ravuri’s earlier days at Qualcomm, where he worked as a vice president of mobile product management. At the time, the technology giant was developing the next generation of mobile connectivity that would build on the 4G capabilities that enabled the rise of “killer apps.” Vinay saw an opportunity with 5G to target a sector that’s bigger than individual smartphone users.

“We enable private installations…in environments where wi-fi doesn’t actually work well: factories, places where the reliability matters a lot. Sometimes it’s schools, sometimes it’s airports, hospitals, things like that,” he said.

Vinay also saw a “springboard” for a new company leveraging the RISC-V ecosystem, the open-source computing language that makes it easier for wireless devices to communicate and interface with each other, as well as the emerging OpenRAN ecosystem that aims to make pieces of radio networks more interchangeable.

When Ravuri started EdgeQ in 2018, he said he saw an opportunity because the market for 5G chips parallels “the old PCs and servers where they were vertically integrated, whether it’s IBM or these other companies,” he said. “There was no real silicon company that can go horizontal.”

In going horizontal, the company has an arrow nocked at merchant silicon vendors like Qualcomm and MediaTek that have traditionally dominated the standalone processor market. Other major players that make chips—namely Huawei, Samsung, Ericsson, and Nokia—typically do so for their own products.

There are obviously a plethora of uses for chips aside from powering advanced connectivity in mobile phones, including Ravuri’s enterprise network targets. EdgeQ isn’t coming for the handset chipmakers’ lunch—at least not just yet.

“[We’re looking at] IoT devices, base stations, access points—the network side of things,” he said. “For a startup to go and just completely replace Qualcomm in these big markets is not easy…We wanted to focus on building this IP, but yet still generate revenue in near-term places that are much easier. We do plan to eventually look at the phone and these [other] high-volume markets as well.”

For now, Ravuri is looking ahead to seeing EdgeQ’s products at work in the real world.

“We just were in production as of six months ago. So this year, we’re hoping to scale, meaning our customers go put [our chips] into their networks and they generate revenue,” he said. “It’s prime time this year.”

Semiconductors

How is tariff uncertainty impacting tech?

Nvidia sought to show AI hype has legs in big earnings beat

US semiconductor manufacturing projected to skyrocket in wake of CHIPS Act

Biden admin announces plan to shell out $285m for digital twins of semiconductors

Nvidia announced its most powerful chip yet at ‘AI Woodstock’

Semiconductors, VR, 5G, AI led new patent grants last year

This 5G chip startup plans to go toe-to-toe with telecom giants. Here’s how

Keep up with the innovative tech transforming business