TheChek News Flow

OpenAI’s GPT-4.5 Underwhelms; Is Poolside Worth $5 Billion?

On X, CEO Sam Altman described GPT-4.5 as a “giant, expensive model” that “won’t crush benchmarks.” By OpenAI’s own admission, the new model generally performs worse than a number of other models, such as Anthropic’s just-launched Claude 3.7 Sonnet and even some of OpenAI’s own reasoning models that have been available for months. It’s also incredibly expensive, costing more than 10 times more than 3.7 Sonnet.

The release received largely unenthusiastic responses on X from developers, though some felt it was better than GPT-4o, which came out in May last year, for some queries.

While GPT 4.5 has redeeming qualities such as more realistic dialogue and a better sense of humor, which could be useful for voice interactions, the model’s performance is clearly nowhere near what the company had originally hoped for. It looks like evidence of slowing gains from pretraining, which we first told you about in November.

The question now is when will OpenAI launch what it calls GPT-5, combined with its o3 reasoning model.

As good as OpenAI is at shipping technology quickly, it also has a tendency to tease new products long before they’re ready. It’s been more than two months since OpenAI announced its performance evaluations of o3, prompting breathless fanfare on X, yet Altman recently said o3 wouldn’t ship as a standalone model the way o1 shipped.

If OpenAI doesn’t launch a model showing meaningful improvement soon, the growing chorus about its eroding lead will get louder.

In other developments…

Even as OpenAI and Anthropic increasingly focus on artificial intelligence products for software engineers, venture capitalists are constantly funding would-be challengers at sky-high prices.

In recent weeks, for instance, investors have been talking to executives at Poolside, a two-year-old AI coding developer founded by one of the creators of GitHub’s Copilot coding assistant, about investing more in the company, according to three people who spoke to Poolside leaders.

The company could target a $5 billion valuation in an eventual fundraising, according to a person who recently spoke to Poolside CEO Jason Warner. (For its part, Poolside isn’t interested in raising new financing now and hasn’t been discussing a potential valuation, a person close to the company said.)

A $5 billion valuation would be striking given that Poolside raised $500 million at a $3 billion valuation in a deal led by Bain Capital Ventures just four months ago. And the company generated less than $10 million in revenue in 2024, according to two of the people who spoke to its executives, though we don’t know what it’s projecting this year.

Poolside is developing a coding assistant app and two coding-focused models—one for more complex engineering tasks and the other for simpler code completion. Though the products aren’t available broadly to developers yet, they will likely compete with similar tools from OpenAI and Anthropic.

Investors such as Vinod Khosla, OpenAI’s first venture investor, are also looking to back startups developing software that essentially aim to replace engineers entirely, such as coding agent Devin, he said at The Information’s AI Agenda Live event last week.

OpenAI has been developing a similar product to Devin. And much of ChatGPT’s subscription revenue comes from software engineers who use the chatbot for coding tasks. That’s one of the reasons investors are going ga-ga for the coding sector.

Even if Poolside hypothetically hoped to generate $100 million in revenue this year, a $5 billion valuation would still rank it among the most expensive generative AI startups to have raised money recently, in terms of valuation multiples.

Even at its existing $3 billion valuation, Poolside looks pricey compared to the three-year-old startup behind AI coding assistant Cursor, which recently hit $100 million in annual recurring revenue—subscription revenue it could generate over the next 12 months—and it has raised money at a $2.5 billion valuation. Codeium, another AI coding assistant startup, is in talks to raise funding at a nearly $3 billion valuation in a round led by Kleiner Perkins, though we don’t have its revenue information.

Homegrown Models

Poolside may get some help from Amazon. In December, the startup announced a deal with Amazon Web Services to sell its coding models—Malibu and Point—as well as its coding product. As part of the partnership, AWS will sell Poolside’s product as a first party service—the cloud giant will make the startup’s product available to customers alongside AWS’ baseline web services like storage and compute—which Poolside expects will help grow its revenue significantly in the coming years, according to one of the people who spoke to the startup.

To sell its product, Poolside is pitching senior executives and decision makers at large companies, according to one of the people. Poolside’s product is tailored to big enterprises: its models are customized on customers’ codebases. Customers can run Poolside’s models on their own servers instead of in the cloud. That’s an important consideration for defense contractors, for instance.

Poolside’s strategy differs from the approach pursued by other AI coding firms, including the makers of Cursor and Devin, which have gained traction with individual software developers.

Unlike many other AI coding-related firms such as the maker of Cursor and even Microsoft’s GitHub Copilot, Poolside doesn’t rely on models from Anthropic and OpenAI. Poolside will have to keep its models competitive with those of the incumbents but some investors may prefer that challenge over the risks of relying on competitors.—Aaron Holmes, Sri Muppidi and Stephanie Palazzolo. Natasha Mascarenhas contributed to this report.

Policy Watch

California State Senator Scott Wiener introduced SB 53, a bill to protect whistleblowers at AI companies and to fund AI chips for researchers and startups.

The bill could continue to evolve as Gov. Gavin Newsom’s AI task force releases its own recommendations for new policies, said one person familiar with the bill. Whistleblower protections and state-funded AI capacity were included in SB47, Wiener’s AI regulation bill last year, which Newsom vetoed after the legislature approved it. Anthropic, for its part, had argued that the whistleblower protections in last year’s bill were too broad.—Rocket Drew

Privacy-focused messaging app Signal has been flying high in the Dutch app stores this past month, often sitting at the top as the most downloaded free app on iOS and Android across all categories, per data from multiple app-tracking platforms such as Sensor Tower.

The app has experienced surges in popularity over the years, often in response to policy changes at rivals like WhatsApp or geopolitical events. That’s because Signal has made a name for itself as a more privacy-friendly option — it’s operated by a not-for-profit foundation (albeit one based in the U.S.) rather than a private business focused on monetizing data. Moreover, Signal tracks minimal metadata.

In 2025, with a new U.S. president empowered by Big Tech’s warm embrace, it’s not surprising that digital privacy tools are having a moment — particularly in Europe, which has attracted President Trump’s ire.

But what’s especially eye-catching this time around is Signal’s prominence in one very specific locale — the Netherlands.

In an interview with Dutch newspaper De Telegraaf last week, Signal President Meredith Whittaker noted that the number of “new registrations” in the Netherlands was higher this year by a factor of 25, though it’s not clear what the exact comparative period of time is for this data.

When asked why the Netherlands has seen such growth, Whittaker pointed to a combination of factors: “Growing awareness of privacy, distrust of big tech, and the political reality in which people realize how vulnerable digital communication can be,” Whittaker said.

Data provided to TechCrunch from app intelligence firm AppFigures charts Signal’s rise in the Netherlands. Per its data, Signal ranked 365th among non-game iPhone apps in the Netherlands on January 1 and did not appear in the top overall apps list. Then, starting around January 5, it began to climb the rankings, reaching the top position by February 2.

Signal has dipped in and out of the lead in the intervening weeks, spending around half of February at the summit — including each day since February 22. Digging deeper into the data, AppFigures estimates that combined downloads across Apple and Google’s app stores totaled around 22,000 in December 2024. This jumped to 99,000 in January and soared to 233,000 through February — a 958% rise since December.

While some of this growth may be attributed to Signal having lower saturation than in other markets, the app’s sustained position at the top compared to similar-sized neighboring markets is notable.

“No other markets come close to the Netherlands in terms of growth between December and February,” AppFigures told TechCrunch.

For comparison, since December, Belgium has seen downloads grow by more than 250%, Sweden by 153%, and Denmark by 95%.

So why might Signal be experiencing what one Redditor called a “mass adoption moment” in the Netherlands?

Clear signal

Rejo Zenger, senior policy advisor at Dutch digital rights foundation Bits of Freedom, said that while it’s difficult to pinpoint one specific reason, he’s not surprised.

Recent developments in the U.S. have seen the big platform providers align with the new Trump administration, and this has stoked significant public and media debate. Europe’s reliance on technology from huge private U.S. companies has become a focal point in that debate.

“The Dutch are, just like many others, highly dependent on the infrastructure provided by extremely dominant tech companies, mostly from the U.S.,” Zenger told TechCrunch. “What this means, and the risks that come from this, have been nicely demonstrated in the past few weeks. As a result, the public debate in the Netherlands has been relatively sharp. Where in the past this problem was only discussed on the level of ‘which instant messenger should I use,’ I feel now we are having the debate on higher levels as well: ‘we should get rid of this dependency.’”

In that context, the public could be conflating dominance with data protection abuse. With companies like Meta regularly being investigated and fined over data privacy practices, Signal might appear the lesser evil: it’s based in the U.S., but operated by a non-profit that ensures encryption of both message content and the metadata around it.

Vincent Böhre, director at Dutch privacy organization Privacy First, also pointed to increased media coverage and a broader shift in public opinion.

“Ever since Trump was re-elected in the U.S. a few months ago, there has been a lot of ‘bashing’ of Trump and [Elon] Musk in Dutch — and European — mainstream media, including bashing of American Big Tech companies, which now seem to be supportive of Trump,” Böhre told TechCrunch. “Articles criticizing X [formerly Twitter] and Meta have been popping up in Dutch media everywhere, leading to a shift in Dutch public opinion: even people who never really knew or cared about privacy and security in social media, have now suddenly become interested in ‘privacy-friendly’ alternatives, Signal in particular.”

Signal of intent

While the Netherlands is just one market of 18 million people in a European population of more than 700 million, its surge in adoption could signal a broader trend across the continent, especially as governments seek to bring down privacy barriers.

Apple, for example, recently pulled end-to-end encryption from iCloud in the U.K. to counter government efforts to install a backdoor.

Speaking at RightsCon 25 in Taiwan this week, Whittaker reaffirmed Signal’s unwavering stance on privacy.

“Signal’s position on this is very clear –- we will not walk-back, adulterate, or otherwise perturb the robust privacy and security guarantees that people depend on,” Whittaker said. “Whether that perturbation or backdoor is called client-side scanning, or the stripping of the encryption protections from one or another features similar to what Apple was pushed into doing in the U.K.”

Separately, in an interview with Swedish public broadcaster SVT, Whittaker said Signal would not comply with a proposed Swedish law requiring messaging app-makers to store messages.

“In practice, this means asking us to break the encryption that is the foundation of our entire business,” Whittaker said. “Asking us to store data would undermine our entire architecture and we would never do that. We would rather leave the Swedish market completely.”

TechCrunch reached out to Signal for comment, but hadn’t heard back at the time of publishing.

Portfolio trading — now also A Thing in European bonds

We love big bond trades and we cannot lie

Yes, we’re going to write about portfolio trading, again. Our previous posts have mostly focused on how a big deal it has become in the US, but a new Barclays report shows it’s now a growing phenomenon in Europe as well.

Barclays has previously estimated that a big fat portfolio trade now hits the US fixed-income market every seven minutes, with volumes topping $1tn annually. Portfolio trading now accounts for as much as a quarter of investment-grade corporate bond trading, and a sixth of junk bond trading.

Calculating the same in Europe is tricky because of the lack of any consolidated tape of bond trades (though that’s about to change) so the UK’s bank’s analysts used an algorithmic approach to identify probable portfolio trades and get a rough estimate.

Barclays found that total annual volumes probably topped €250bn in 2024. In frequency, that equates to a portfolio trade happening every 20 minutes on average, down from every 33 minutes just two years ago.

It has become particularly prevalent in investment grade, euro-denominated corporate bond market, where portfolio trading now accounts for 11 per cent of all trading volume. In sterling bonds and euro high yield it has reached a 9 per cent and 7 per cent market share respectively.

Most portfolio trades identified by Barclays happened on electronic platforms like Tradeweb, and involved about 60 bonds worth a total of €40mn on average.

That’s lower than the $50-60mn average in the US, but this is maybe less than you you’d expect, given how much larger and more liquid the US corporate bond market is, and how developed portfolio trading is becoming there.

Zornitsa Todorova, head of thematic fixed income research at Barclays, reckons that portfolio trading volumes in Europe will roughly double to account for about 20 per cent of all dealer-to-client volumes within the next three years.

How big a deal is all this? Well, pretty big if you’re a bond fund manager.

Index-tracking fixed income vehicles often use portfolio trades to rejig their exposures, but it has become a hugely valuable tool for active managers as well, as Todorova highlights. Her emphases below:

Our analysis attributed the majority of portfolio trading volumes in Europe to active investors, though we found that passive investors do use PTs to rebalance portfolios at month-end and to manage inflows and outflows intra-month by buying or selling a ‘slice’ of their holdings that closely resembles the index they track. In addition to these use cases, active managers also frequently use PTs to adjust their portfolios towards specific credit risk dimensions, such as rating, sector or maturity, along with hedging and de-risking.

The fact that active investors are the marginal consumers of PT liquidity is significant for several reasons. Unlike passive investors, who track an index, active investors, in particular open-end mutual funds, frequently trade bonds and depend on liquidity to efficiently enter and exit positions. These investors set marginal prices when they trade, influencing bond spreads and market depth. Their activity also impacts liquidity risk premia, as bonds with lower liquidity command higher yields.

Portfolio trading has the potential to transform credit markets in several ways: by adding more volume to the market, it improves liquidity, lowers the barrier to entry for new investors, and enables new strategies, such as systematic credit.

The evidence is clear that buy and sell recommendations should be treated with caution

Analyst ratings on stocks are a regular feature in financial news coverage. Upgrades, downgrades and price target changes can often create volatile one-day price movements and attract investor attention.

However, investors should be wary of these headline-grabbing stock reports from experts at large firms. The long-term investment implications might not be what they expect. In fact, following sell-side analyst advice may lose you money.

My firm analysed the investable information to be gleaned from analyst ratings, changes in ratings and price targets. Is there much value in these? The short answer is — there is not.

The S&P Capital IQ analyst rating summary score for the sell-side analyst recommendations is between one and five and measures how loved or hated a stock is by the sell-side community. We found that the best-performing stocks over the past 25 years are those in the lowest quintile.

In addition, over the past 24 years, if you bought the stocks rated most highly by the sell-side analysts and shorted the stocks with the lowest number of buy ratings, you would have lost money — about 30 per cent cumulatively, with meaningful volatility.

For one example of how following buy ratings doesn’t always work best, take Microsoft. On October 1 of last year, there were 53 buy, three hold and zero sell recommendations by the sell-side. The stock has lagged the S&P 500 by more than 10 percentage points since.

The evidence is clear that investors should not look for investment ideas based on a high percentage of buy recommendations from sell-side analysts.

What about change in a data point, not the level of it? That is often more important in equity analysis. A fair follow-up question could be: what if analysts change their minds?

If something is shifting enough to induce a change to an analyst’s rating, that outlook might matter. Hence, we studied the month-over-month change in analyst ratings and found that this metric was not an effective predictor of subsequent return. Over the past 20 years, stocks recently upgraded by analysts did not outperform those with recent downgrades.

In addition to using the S&P Capital IQ score, we used another common institutional investor approach, looking at the buy, hold and sell ratings on Bloomberg. We defined “buy percentage” as the number of analysts with a “buy” recommendation on a stock divided by the total number of analysts covering a stock. We then looked for stocks newly in favour and those “incrementally loved”, as indicated by an increased number of buy ratings, and compared their subsequent stock performance with those with the most month-over-month downgrades.

These “incrementally loved” stocks did beat the incrementally disfavoured ones from 2001 through 2013 but have generated no additional return over the past dozen years. Recently upgraded stocks do not outperform recently downgraded stocks, so looking for changes in rating is also not a fruitful investment discipline.

As for price targets, there is a notion that if the analysts see tremendous upside to a stock, they set price targets way above where a stock is trading as a signal of their confidence that the security will appreciate. Should investors forget about ratings and changes in them, instead of just looking for stocks forecast to have big gains?

That is also not a good strategy. Stocks with lots of “upside” from the current price to the median price target might be perceived as ones where analysts are bullish. But they are often stocks that analysts had been previously too bullish as indicated by bad price momentum.

The outlook for a stock that has been down recently can be meaningfully different for one trading at highs, even if they both have lofty price targets set by analysts. Stocks with high price targets performed best from 2009 to 2016, but the signal has diminished over the past decade.

We did find modest value in the volatility of the price targets for an individual stock. Companies with the lowest standard deviation around their analyst price targets have tended to outperform those with high variability in targets. However, the metric peaked in its efficacy in September of 2022 and has not been of much use since.

But overall, relying on analyst ratings and rating changes has not been a good strategy historically — and is not today. Sell-side analysts, of course, offer insightful original industry research and quantitative analysis that is valuable to investors. But we would suggest looking beyond the eye-catching ratings and price targets.

Hayden Davis had a string of short-lived business ventures when he met Javier Milei. Together they pumped a crypto token that’s shaking Milei’s presidency.

He is a college dropout who hawked energy drinks as a teen and sold Oreos to make rent. Under “skills” on LinkedIn, he listed “hustling expert.”

Last month, Hayden Davis, 28 years old, became the face of the highest-profile cryptocurrency scandal since the implosion of FTX in 2022, one that has ensnared Argentine President Javier Milei.

Propelled by self-confidence and networking savvy, Davis went from cookie seller to erstwhile crypto king, according to interviews and a review of more than a decade of social-media posts, podcasts and blog posts.

Davis, his family and their venture-capital firm, Kelsier Ventures, were behind the Feb. 14 launch of $LIBRA, a meme coin that was intended to help fund the development of the Argentine economy, according to a website describing the project. When Milei promoted $LIBRA on social media, the price soared—until it crashed, eviscerating up to roughly $1 billion in market value, according to DEX Screener, a crypto data provider.

More than 10,000 investors have lost a total of at least $250 million, according to blockchain analytics firm Nansen. Davis, though, has acknowledged that he walked away with close to $100 million by quickly selling before the crash.

Davis, who didn’t respond to numerous requests for an interview, said in podcast interviews after $LIBRA’s crash that while he regretted that investors had lost so much, meme coins are inherently hazardous.

Milei didn’t respond to a request for comment.

Meme coins are a type of digital token created often as jokes and mainly for speculation. As news of a token launch spreads on social media, meme coins can spike in value until a small group of investors—often insiders—sell at a profit, saddling the remaining investors with losses.

$LIBRA is only the latest example of presidents joining the meme-coin craze. In January, President Trump launched a coin, $TRUMP, that also saw a massive price spike followed by a rapid fall. Days later, first lady Melania Trump released her own version, $MELANIA, a project Hayden Davis has said he was also involved in.

Kelsier is led by Davis’s father, Tom Davis, a felon who led an international Christian adoption nonprofit, then became a leadership coach and serial entrepreneur, according to law-enforcement and nonprofit records, social-media posts and his public statements. Through Hayden’s mother, the family springs from a line of former adherents to a violent Christian fundamentalist sect, the Church of the First Born of the Lamb of God. Her father was killed by believers after fleeing the group when she was a young child, she said on an episode of her sons’ podcast. She was never a member of the sect.

Now threats of lawsuits and demands for criminal investigations are circling Milei and Davis. Argentina’s federal prosecutor has opened a fraud investigation into the launch of $LIBRA, citing allegations that crypto entrepreneurs paid for access to Milei.

A smooth talker

By the time he launched $LIBRA, Davis had a string of short-lived business ventures more than a decade long.

He started selling energy drinks for Limu, a multilevel marketing company his father was involved in that sold beverages including Blu Frog and Lean Burn, when he was 17.

He attended the evangelical Christian college Liberty University for two semesters on a soccer scholarship, a spokesperson for the school confirmed. He launched a T-shirt printing company, dabbled in what he described on a podcast as “the world of e-commerce,” interned for a private-equity firm, and briefly joined a professional Spanish soccer team.

By 2021, he was living in Los Angeles with two roommates, in the middle of what he referred to as his “fifth business failure” on a podcast he hosted with his brother Gideon.

That is when the Davis family began embracing cryptocurrency. Lured by tax advantages the city had offered the industry, Tom Davis began doing business in Dubai under the name Kelsier.

Kelsier began making early-stage investments, including the cryptocurrency exchange Meteora, where both $LIBRA and $MELANIA launched. Later it offered marketing services to crypto exchanges, nonfungible tokens and meme coins, according to founders who worked with Kelsier.

At the time, Tom Davis was far from a cryptocurrency expert. Earlier he had gotten into trouble forging checks, he told the Christian Broadcasting Network, and pleaded guilty to burglary, according to law-enforcement records. He went to Bible school and became an entrepreneur, investing in bricks-and-mortar businesses such as Skrimp Shack, a chain of seafood restaurants in the U.S. and the Middle East.

Though Tom Davis was Kelsier’s founder, Hayden was its face. Industry peers who worked with Hayden said he’s a smooth talker and a natural dealmaker.

“If you sit down with him and he wants to sell you something, he will convince you,” said Steven Enamakel, a Dubai-based crypto entrepreneur who briefly worked with Hayden on a project unrelated to $LIBRA.

Hayden also had a knack for getting into high-profile projects. Ben Chow, the co-founder of Meteora, said in a social-media post that he referred Hayden and Kelsier to the team behind the $MELANIA token because “they appeared to be trustworthy.”

Chow, who recently resigned from Meteora, declined requests for an interview.

By the fall, Hayden and Gideon were helping to broker a deal between Milei and the crypto exchange Cube, a trading platform founded by an ex-developer from Ken Griffin’s hedge fund Citadel to “help with crypto stuff over there,” Gideon said on a podcast.

“It’s signed and stamped by Javier Milei,” he said on the podcast. “We’re doing an event with him in two weeks.”

In October, Hayden, Tom and Gideon Davis sat next to Cube’s founder as Milei gave a speech at a technology conference organized by Argentine entrepreneur Mauricio Novelli, according to a video of the event. The next month, Hayden Davis and Novelli were spotted entering Argentina’s presidential palace, local media reported.

In a statement, Novelli said he didn’t handle $LIBRA funds or make any money trading the coin.

‘LONG LIVE FREEDOM, DAMN IT…!!!’

By January, Hayden was riding high.

The $MELANIA coin launched on Jan. 19, soaring to a peak market cap of about $2 billion before falling, though Hayden said in an interview with the crypto journalist and YouTuber Stephen Findeisen, known online as Coffeezilla, that he made “no money” on the coin.

(A White House spokesperson referred questions to the Trump Organization, which didn’t respond to a request for comment.)

On Jan. 30, Hayden got a bump from Milei, who touted their connection, posting a selfie of them to X and writing that Hayden was “advising me on the impact and applications of blockchain technology and artificial intelligence in the country.”

“LONG LIVE FREEDOM, DAMN IT…!!!” Milei added.

Two weeks later, Davis and his partners released $LIBRA. The evening of the launch, Milei posted on X in support of $LIBRA, calling it a project that will “focus on encouraging the growth of the Argentine economy.” Within hours, the token surged—and then lost most of its value by midnight.

Milei deleted his original post, claiming in a new post he hadn’t been “aware of the details of the project.” But in a video posted to X the day after the launch, Hayden appeared to underscore his relationship with the president.

“I am indeed Javier Milei’s adviser,” Davis said.

By Sunday, Feb. 16, the Argentine president was facing backlash from investors, calls for impeachment from his political opponents and fraud complaints from lawyers. An Argentine law firm asked the Federal Bureau of Investigation and the Securities and Exchange Commission to open an investigation into Hayden Davis. Both agencies declined to comment on the existence of an investigation.

In later interviews, Hayden defended the practice of using insider knowledge to make money from tokens such as $MELANIA and $LIBRA.

“It is an insider’s game,” he said in the interview with Findeisen. “This is an unregulated casino.”

Milei, too, compared the crypto market to gambling in an interview with Argentine media outlet Todo Noticias, adding that people who lost money on $LIBRA had no right to complain.

“It’s like someone who goes and plays Russian roulette and the bullet hits him,” Milei said.

Still, Hayden Davis has expressed remorse for the way $LIBRA turned out. In media interviews, he’s contemplated using the $100 million he made—money he has said he believes belongs to Argentina—to prop up the price of $LIBRA. And he’s said that he has done nothing different from any other meme-coin founder.

“Maybe I shouldn’t have done that in hindsight, whatever,” he told Barstool Sports founder Dave Portnoy in a livestream after the scandal broke. “But I’m not a piece of s—.”

Canada and Mexico levies would hit hardest among slew of Trump trade measures that could raise vehicle prices

After a month’s reprieve, the U.S. autos industry is again staring down 25% tariffs on imports from Mexico and Canada, levies that would inject costs into the industry’s highly integrated North American supply chain.

President Trump has said the tariffs would take effect Tuesday. The Canada and Mexico levies, if fully applied to vehicles and car parts, are the most immediate of several tariff scenarios facing the industry.

Trump also is planning to add 25% tariffs on imports of steel and aluminum, key materials in vehicles, on March 12. The president has also floated “reciprocal tariffs” to match duties of other countries and sector-specific tariffs that would include the auto industry. Tariffs on Chinese imports also are headed higher.

The umbrella of tariffs would raise costs for foreign- and U.S.-built vehicles, because even domestic cars include a significant amount of parts and content from abroad. The levies on Mexico and Canada alone could add around $3,000 to the average U.S. car price, analysts estimate.

Here is a look at the industry’s North American trade landscape:

Slightly more than half of U.S.-sold vehicles are built in the U.S.

Mexico is the largest importer of finished vehicles to the U.S., including many popular models by U.S. manufacturers such as the Chevrolet Equinox, Ford Bronco Sport and some Ram pickups. The other major foreign sources of vehicles are South Korea, Japan, Canada and Germany.

Some carmakers rely heavily on Mexico, Canada for U.S. sales

Nearly a quarter of new passenger vehicles sold in the U.S. in 2024 were assembled in Mexico or Canada. Carmakers with above-average reliance on the two countries include Volkswagen VOW3 4.37%increase; green up pointing triangle, Stellantis STLA 1.27%increase; green up pointing triangle, Honda 7267 1.40%increase; green up pointing triangle and General Motors GM 3.15%increase; green up pointing triangle.

U.S.-built vehicles have lots of foreign parts

Tesla TSLA 3.91%increase; green up pointing triangle, run by Trump ally Elon Musk, builds all of its U.S.-sold cars domestically, but those cars include about 20% of parts and components from Mexico, according to a U.S.-mandated auto labeling program overseen by the National Highway Traffic Safety Administration.

Federal auto-labeling requirements show that a significant percentage of parts in some popular U.S. vehicle models originate outside the U.S.

New vehicles are already expensive

Tariffs would add costs at a time when U.S. vehicle prices remain elevated after surging during the Covid-19 pandemic. The average price of about $44,000 is up 25% since 2019.

The Canada and Mexico tariffs would add an average of $3,125 in costs per vehicle, analysts at JP Morgan said in a research note. It isn’t clear what share of those costs would be absorbed by automakers and their suppliers versus ultimately passed on to car buyers.

Another key question is whether the administration will include any exemptions for vehicles that comply with Trump’s free-trade agreement, the U.S.-Mexico-Canada Agreement.

Gapping down
In reaction to earnings/guidance:

Other news: