semiconductor

Published on 29/10/2025 – 11:09 GMT+1
•Updated
11:11

Nvidia shares continued their dramatic rise this week as investors banked on an easing of semiconductor trade restrictions between the US and China.

Ahead of a meeting with Chinese President Xi Jinping on Thursday, US President Donald Trump said he planned to discuss Nvidia’s advanced Blackwell artificial intelligence chip with Xi.

“We’ll be speaking about Blackwell, it’s the super duper chip,” he told reporters on Wednesday.

The president didn’t elaborate on specific policy aims, although he said he was “very optimistic” about the meeting with his Chinese counterpart.

By around 11:00 CET, Nvidia shares had jumped over 3% in pre-market trading, bringing the firm closer to a $5 trillion market capitalisation.

Semiconductors have been a key point of contention between the US and China as both nations seek to lead on advanced technologies such as AI.

The tiny chips, used to power a range of electronic devices from smartphones to medical equipment, are essential to this ambition. Since 2022, the US has therefore restricted Nvidia’s sales of advanced chips to China for national security reasons.

Trump has flip-flopped on export controls since his arrival in the White House, first restricting and then approving sales of Nvidia’s H20 AI chip to China. Nvidia designed the H20 specifically for the Chinese market to comply with Biden-era export curbs, although the Trump administration previously said it was concerned the tech could be used for military purposes.

With regard to the Blackwell processor, Trump suggested months ago that he would consider allowing Nvidia to export a downgraded version of the chip to China.

Progress on such a proposal would come as a relief to Nvidia CEO Jensen Huang, who has long criticised US restrictions. Huang has notably argued that such curbs are boosting China’s AI capabilities as the Chinese market is forced to become less reliant on US products.

It seems that such logic is already understood in Beijing, even as the US softens its stance. After Washington gave the green light to H20 exports, China’s regulator banned the country’s biggest tech companies from buying Nvidia’s artificial intelligence chips.

“The president has licensed us to ship to China, but China has blocked us from being able to ship to China,” Huang said at a Nvidia event this week in Washington. “They’ve made it very clear that they don’t want Nvidia to be there right now.”

In a document released by Beijing on Tuesday, the Communist party reiterated the importance of self-sufficiency, calling for “extraordinary measures” to achieve “decisive breakthroughs” in technologies such as semiconductors.

“The most important factor in promoting high-quality development is to accelerate high-level scientific and technological self-reliance,” Xi said in a speech released by state news agency Xinhua.

While it’s possible that Chinese restrictions on Nvidia chips could be a long-lasting policy, experts have suggested that the move may be a bargaining chip in trade negotiations with Washington.

Such policy U-turns are creating uncertainty for investors despite the fact that Nvidia shares have risen roughly 50% this year, driven higher by AI ambitions.

Nvidia has been the dominant force in the global semiconductor industry thanks to its graphics processing units (GPUs), which have played a critical role in enabling the proliferation of artificial intelligence (AI) applications. The demand for Nvidia’s GPUs has been so solid in the past three years that Nvidia has now become the world’s largest company.

Nvidia continues to rule the AI data center GPU market, facing very little threat from its peers so far. Analysts are expecting its top line to jump by an impressive 58% in the current fiscal year to more than $206 billion. That’s quite impressive for a company of Nvidia’s size. The stock registered respectable gains of 34% on the market this year based on the healthy growth that the company continues to deliver.

However, Nvidia’s stock market performance has been overshadowed by Broadcom (AVGO -1.24%). Broadcom has appreciated 48% this year and looks set to end 2025 on a high note following recent developments. In fact, it won’t be surprising to see Broadcom stock outperforming Nvidia next year as well. Let’s see why that may be the case.

Image source: Nvidia.

Custom AI chips are expected to witness stronger demand in 2026

So far, the majority of AI model training and inference has been carried out by Nvidia’s GPUs. GPUs are general-purpose computing chips with massive parallel computing power, making them ideal for quickly training AI models and moving them into production. OpenAI chose Nvidia’s A100 data center GPUs to train its popular chatbot ChatGPT three years ago.

Nvidia built upon its first-mover advantage and controlled an estimated 92% of the AI data center GPU market at the end of last year. However, the latest deal struck between OpenAI and Broadcom indicates that Nvidia’s influence over the AI chip market could wane. OpenAI will buy custom AI accelerators worth a whopping 10 gigawatts (GW) from Broadcom starting in the second half of 2026.

The deployment is expected to be completed by the end of 2029. This is a massive deal for Broadcom considering that it reportedly costs around $10 billion to build a 1 GW data center. Around 60% of the investment that goes into building a data center is allocated toward chips and other computing hardware, which would put Broadcom’s potential addressable market from each gigawatt of OpenAI’s deployment at $6 billion.

So, Broadcom could be sitting on a potential revenue opportunity worth $60 billion from this deal over the next three years. Broadcom’s custom AI processors have already been in terrific demand as hyperscalers and AI giants such as OpenAI are gravitating toward these chips because of the advantages they enjoy over GPUs.

Custom AI processors are designed for performing targeted tasks, such as AI inference. As a result, they are not only more power-efficient at running those workloads but also enjoy a performance advantage since they don’t need to perform any other tasks. Hence, deploying custom AI processors can help save costs for hyperscalers.

Shipments of application-specific integrated circuits (ASICs) meant for deployment in AI data centers are expected to increase by 45% in 2026, compared to the expected growth of 16% in GPUs. Broadcom is in the best position to make the most of this growth opportunity as it leads the ASIC market with an estimated share of 70%.

Moreover, the new deal with OpenAI along with another $10 billion contract with an unnamed customer that the company announced last month should ensure outstanding growth in Broadcom’s AI revenue next year.

Broadcom’s AI revenue could now increase at a faster pace

Broadcom is on track to end the current fiscal year with almost $20 billion in AI revenue, an increase of 64% from the previous year. The company reported a record revenue backlog of $110 billion at the end of the fiscal third quarter (which ended on Aug. 3). That backlog is likely to have moved higher following the recent deals struck by the company.

Don’t be surprised to see Broadcom’s revenue jumping at a faster pace than the 33% growth that Wall Street is expecting next fiscal year, which would be a nice improvement over the 23% growth it is expected to deliver in the current one. There is a good chance that its revenue growth in the long run could be better than expectations as well.

AVGO Revenue Estimates for Current Fiscal Year data by YCharts

Broadcom was already anticipating a serviceable addressable market worth $60 billion to $90 billion based on the three AI customers it was serving until earlier this year. That addressable market is now much bigger following the OpenAI contract, which opens up the possibility of stronger growth and more upside for Broadcom investors.

Nvidia (NVDA 1.04%) has hit it out of the park quarter after quarter when reporting earnings, but that hasn’t made investors blasé about the artificial intelligence (AI) giant’s next update. Instead, investors wait with just as much anticipation each time around — and even wonder if, this time, they’ll see a slowdown in what’s been a whirlwind growth story.

As investors count the days until the next report — and in this case, it’s set for Nov. 19 — they look for clues about Nvidia’s AI business, one that’s generated record revenue in recent years. Nvidia, as the world’s biggest AI chip designer, delivered $130 billion in revenue in the latest fiscal year — that’s compared to $27 billion just two years earlier.

Now, one particular clue — and one investors truly can count on — comes from Taiwan Semiconductor Manufacturing (TSM -1.68%), a key Nvidia partner. TSMC, the world’s largest chip manufacturer, just announced big news for Nvidia stockholders. 

Image source: Getty Images.

How Nvidia and TSMC work together

Before we get to this fantastic news, though, we’ll take a quick look at Nvidia’s business and how the company works with TSMC. Nvidia for many years built its business around designing chips for the gaming market, but as AI surfaced as a growth opportunity, the company turned its attention there. And, as they say, the rest is history.

Today, Nvidia dominates this market with its high-powered chips as well as related products and services from enterprise software to networking systems. This has helped earnings and the stock price soar — Nvidia shares have climbed more than 1,100% over the past five years.

It’s important to note that though Nvidia is a chip designer, it’s not a chipmaker. Nvidia doesn’t actually manufacture its AI chips, known as graphics processing units (GPUs), and instead turns to TSMC for that job. TSMC has more than 500 customers across segments of the market, including the world’s chip leaders — from Nvidia to Broadcom and Advanced Micro Devices.

A deep look at the industry

On top of this, since the actual production of advanced chips becomes more and more complex with each chip innovation, TSMC starts work with customers two to three years prior to a new project. “Therefore, we probably get the deepest and widest look possible in the industry,” CEO C.C. Wei said during the company’s earnings call this week.

All of this means TSMC has a very clear picture of what’s happening in today’s AI market and what lies ahead. And this brings me to the news the company delivered this week — news that’s a big deal for Nvidia stockholders.

TSMC reported a 39% increase in profit and a 30% increase in revenue in the recent quarter, beating analysts’ estimates. Importantly, Wei said TSMC continues to see a “strong outlook” from customers and “received very strong signals from our customers’ customers. … Our conviction in the AI megatrend is strengthening.” Wei added that semiconductor demand “will continue to be very fundamental.”

Confirming the trend

All of this is incredible news for Nvidia’s shareholders as it confirms the trends the chip designer has spoken of in recent quarters and its prediction for growth in demand. In Nvidia’s most recent earnings report, back in August, CEO Jensen Huang predicted that AI infrastructure spending may jump to $4 trillion by 2030. TSMC’s report this past week offers us reason to be optimistic about that possibility and suggests that Nvidia is already starting to reap the rewards.

As customers seek GPUs, chip designers must turn to TSMC for production — and it’s likely that TSMC’s revenue gains reflect demand for Nvidia’s chips since Nvidia is the market leader.

All of this means there’s reason for investors to be optimistic about Nvidia’s upcoming earnings report and the messages it will deliver regarding future demand for its GPUs. That’s incredible news for Nvidia stockholders — and makes the stock a great one to buy and hold today.

The artificial intelligence (AI) revolution is transforming every corner of the global economy. Nvidia, the company at the center of this revolution, continues to be a Wall Street favorite for all the right reasons. As an undisputed leader in accelerated computing, the company’s hardware and software power much of the world’s AI infrastructure buildout.

Shares of Nvidia have already surged over 43% so far in 2025. However, despite the massive demand for its Blackwell architecture systems, software stack, and networking solutions, the stock may grow quite modestly in future months. With its market capitalization now exceeding $4.6 trillion and shares trading at a premium valuation of nearly 30 times forward earnings, much of the optimism is already priced in.

Memory giant Micron (MU 6.12%), on the other hand, is still in the early stages of its AI-powered growth story. Shares of the company have surged nearly 128% in 2025, which highlights the increasing investor confidence in its high-bandwidth memory and data center portfolio. Yet, Micron could still offer investors higher returns in 2026, while riding the same AI wave. Here’s why.

Image source: Getty Images.

Lower customer concentration risk

Wall Street has been highlighting one significant underappreciated risk for Nvidia. Nvidia’s revenues depend heavily on a few hyperscaler customers, with two accounting for 39% and four accounting for 46% of its revenues in the second quarter of fiscal 2026 (ending July 27, 2026). Many of these hyperscaler clients are developing proprietary chips, which may offer a price-performance optimization in their specific workloads. This may reduce their dependence on Nvidia’s chips in future years.

Micron’s revenue base is significantly more diversified than Nvidia’s. The company’s largest customer accounted for 17% of total revenue, while the next largest contributed 10% in fiscal 2025 (ending Aug. 28, 2025). The company has earned over half of its total revenues from the top 10 customers for the past three years. The company has a reasonably broad customer base, including data center, mobile, PC, automotive, and industrial markets.

Hence, compared with Nvidia, Micron’s lower concentration risk makes it more resilient in the current economy.

HBM demand and AI memory leadership

Micron’s high-bandwidth memory (HBM) products, known for their superior data transfer speeds and energy efficiency, are being increasingly used in data centers. HBM revenues reached nearly $2 billion in the fourth quarter of fiscal 2025, translating into $8 billion annualized run rate.

Management expects Micron’s HBM market share to match its overall DRAM share by the third quarter of fiscal 2025. The company now caters to six HBM customers and has entered into pricing agreements covering most of the 2026 supply of HBM third-generation extended (HBM3E) products.

Micron has also started sampling HBM fourth-generation (HBM4) products to customers. The company expects the first production shipment of HBM4 in the second quarter of calendar year 2026 and a broader ramp later that year.

Beyond HBM, Micron’s Low-Power Double Data Rate (LPDDR) memory products are also seeing strong demand in data centers. The data center business has emerged as a key growth engine, accounting for 56% of Micron’s total sales in fiscal 2025.

Hence, Micron seems well-positioned to capture a significant share of the AI-powered memory demand in the coming years.

Valuation

Micron appears to offer a stronger risk-reward proposition than Nvidia, even in the backdrop of accelerated AI infrastructure spending. The company currently trades at 12.3 times forward earnings, significantly lower than Nvidia’s valuation. Hence, while Nvidia’s premium valuation already assumes near-perfect execution and continued dominance, Micron still trades like a cyclical memory stock. This disconnect leaves room for modest valuation expansion to account for Micron’s improving revenue mix toward high-margin AI memory products.

Wall Street sentiment is also increasingly positive for Micron. Morgan Stanley’s Joseph Moore recently upgraded the stock from equal-weight or neutral to overweight and raised the target price from $160 to $220. UBS has reiterated its “Buy” rating and increased the target price from $195 to $225. Itau Unibanco analyst has initiated coverage for Micron with a “Buy” rating and target price of $249.

Analysts expect Micron’s earnings per share to grow year over year by nearly 100% to $16.6 in fiscal 2026. If the current valuation multiple holds, Micron’s share price could be around $204 (up 6% from the last closing price as of Oct. 9), with limited downside potential. But if the multiple expands modestly in the range of 14 to 16 times forward earnings, shares could fall in the range of $232 to $265, offering upside of 20% to 37.8%.

On the other hand, there remains a higher probability of valuation compression for Nvidia, leaving less room for growth. With diversified customers, increasing AI exposure, and reasonable valuation, Micron may prove to be the better semiconductor pick in 2026.

On October 7, 2025, Moulton Wealth Management, Inc disclosed a new position in VanEck Semiconductor ETF(SMH 2.68%), acquiring 8,932 shares valued at approximately $2.92 million.

What happened

According to a Securities and Exchange Commission (SEC) filing dated October 7, 2025, Moulton Wealth Management, Inc disclosed a new position in VanEck Semiconductor ETF, adding 8,932 shares. The estimated transaction value was approximately $2.92 million. The fund reported 45 total positions and $137.49 million in reportable U.S. equity assets.

What else to know

This is a new position; SMH now accounts for 2.1% of the fund’s 13F assets under management.

Top holdings after the filing:

• SPLG: $12.93 million (9.4% of AUM)
• USFR: $10.40 million (7.6% of AUM)
• TFLO: $10.37 million (7.5% of AUM)
• SJNK: $9.82 million (7.1% of AUM)
• FLOT: $9.73 million (7.1% of AUM)

As of October 7, 2025, shares were priced at $337.05, up 35.79% over the past year.

Company overview

Company snapshot

The investment strategy seeks to replicate the performance of the fund’s benchmark index by investing at least 80% of assets in U.S. exchange-listed semiconductor companies.

The portfolio is concentrated in common stocks and depositary receipts of semiconductor companies, including both domestic and foreign issuers.

Fund structure is non-diversified with a passively managed approach.

VanEck Semiconductor ETF (SMH) provides targeted exposure to the semiconductor sector by tracking a benchmark index of leading U.S.-listed semiconductor companies. The fund’s substantial asset base and focused portfolio offer investors a liquid and efficient vehicle for accessing this critical technology industry.

Foolish take

I’m a longtime bull on the VanEck Semiconductor ETF (SMH) for one very simple reason: Semiconductors are a critical component within the artificial intelligence (AI) ecosystem, and AI is the most important technological innovation of this decade.

Therefore, this fund’s core holdings read like a who’s who of top-performing stocks. There’s Nvidia, Advanced Micro Devices, Broadcom, Taiwan Semiconductor Manufacturing, Intel, and many more.

Obviously, many of these stocks have soared to new heights as the AI revolution has picked up steam. Nvidia is now the world’s largest company by market cap; Broadcom is now the 7th-largest American company with a market cap north of $1.6 trillion.

What’s more, organizations are still spending tens of billions on new AI infrastructure investments — much of it coming in the form of purchases of semiconductors.

For example, according to estimates compiled by Yahoo Finance, Nvidia’s annual sales should rise to over $200 billion this year, up from $26 billion in 2022.

All that said, semiconductors have historically been a cyclical industry, and have endured many boom-bust cycles. So investors should remain cautious about how much exposure they may have to the semiconductor industry, given its volatile history.

However, for most growth-oriented investors, semiconductors are now a must-own sector. So for those investors, the Van Eck Semiconductor ETF is one fund to consider for the long term.

Glossary

ETF (Exchange-Traded Fund): An investment fund traded on stock exchanges, holding assets like stocks or bonds.

13F assets under management: The value of U.S. equity securities reported by institutional managers in quarterly SEC filings.

New position: The initial purchase of a security or asset not previously held in a portfolio.

Benchmark index: A standard index used to measure the performance of an investment fund or portfolio.

Depositary receipts: Negotiable certificates representing shares in a foreign company, traded on local stock exchanges.

Non-diversified fund: A fund that invests a large portion of assets in a small number of issuers or sectors.

Passively managed: An investment approach that aims to replicate the performance of a benchmark index, not outperform it.

Expense ratio: The annual fee expressed as a percentage of assets, covering a fund’s operating costs.

Asset base: The total value of assets held by a fund or investment vehicle.

Reportable position: A holding that must be disclosed in regulatory filings due to its size or regulatory requirements.

Taiwan Semiconductor (TSMC) (TSM 1.50%) will release earnings for the third quarter of 2025 on Oct. 16. The company produces the majority of the world’s most advanced semiconductors. Since many of the advancements in artificial intelligence (AI) are not possible without its manufacturing capabilities, the stock is likely to remain a market beater over the long term.

Nonetheless, TSMC stock is at an all-time high, and anticipated growth is often not enough of a reason to buy a stock. With an earnings report looming, should investors buy shares of the stock now or stay on the sidelines and hold out for a lower price?

The bull case in TSMC stock

As previously mentioned, TSMC faces a few threats to its long-term bull case. It is the world’s largest semiconductor foundry company, and as of the second quarter of 2025, its market share now exceeds 70%, according to TrendForce. This is up from 67% in the previous quarter.

Additionally, Grand View Research forecasts a compound annual growth rate (CAGR) for AI of 32% through 2033. These combined factors make it highly likely that TSMC’s rapid growth will continue.

For now, it has exceeded that growth rate, and that rapid growth is on track to continue. In the first half of 2025, revenue increased by 40% to $56 billion compared to the same period the previous year. It also stated on its Q2 earnings call that it expects between $31.8 billion and $33 billion in revenue during Q3, representing a 38% rise at the midpoint.

Investors should note that the company beat revenue estimates in each of the previous four reports. Thus, if it beats estimates like it has in previous quarters, the 40% revenue growth rate from the first two quarters of the year could continue into Q3.

Moreover, investors should watch for sales of the most advanced chips, namely those in the 2nm – 5nm size range that power the most advanced AI functions. This is the area where TSMC stands out above competing foundries, since Samsung is the only other chip producer that can manufacture these smaller chips.

Areas of danger

Additionally, even if it is likely to beat earnings estimates, TSMC faces significant challenges.

One is simply keeping up with demand. It allocated almost $20 billion to capital expenditures (CapEx) in the first half of the year, and much of that will go to foundries in Arizona, where it plans to allocate $165 billion to building six advanced manufacturing facilities. Even though that is a considerable sum, it will likely have to maintain or increase that spending to match demand.

Another factor is that the majority of production takes place in Taiwan, which faces considerable geopolitical tensions because of its proximity to China. Investors differ on the danger level, as China can probably not afford to have the supply of chips disrupted by geopolitical events.

Still, investors should also remember that Warren Buffett forced Berkshire Hathaway to sell its TSMC stake for this reason. Hence, investors must remain aware of this concern.

That issue may also be the reason for TSMC’s relatively low valuation. It has traded at an average P/E ratio of 25 over the last five years, far below its key clients such as Apple and Nvidia.

Also, while its current 33 P/E ratio is low for a company with 40% revenue growth, the earnings multiple has rarely exceeded 40 in recent years. That could increase the danger of paying a relative premium for TSMC.

Should investors buy TSMC stock before earnings?

Under current conditions, no obvious factor is pushing investors to either delay or accelerate purchase decisions before the earnings report.

Indeed, nobody knows how TSMC stock will react once the company releases Q3 earnings. Still, some risk-averse investors may feel apprehensive about the report amid the rising P/E ratio.

If that is the case, one strategy is to do both, allocate half of one’s funds to this stock now and wait for the report to spend the additional half. Shareholders who dollar-cost-average into this stock are likely already employing this strategy, and with this near-term outcome unknown, that approach could also work for other investors.

Ultimately, barring the aforementioned geopolitical risks, TSMC stock should remain on a bull trend as it struggles to meet the demand for AI chips. For this reason, time in TSMC is almost certainly more critical to winning with the stock than the timing of one’s purchase decisions.

When investors think about powerhouses in the semiconductor industry, the usual names that dominate the conversation are Nvidia, Advanced Micro Devices, and Broadcom. These companies are responsible for designing the high-performance chips and networking hardware powering next-generation data centers at an unprecedented scale.

Operating more quietly in the background, however, is Taiwan Semiconductor Manufacturing (TSM -1.17%). While TSMC (as it is also known) is less flashy than its peers in the race for artificial intelligence (AI) chips, the company’s supporting role is nonetheless mission-critical.

As the world’s largest chip foundry by revenue — with almost 70% market share — TSMC is the manufacturer behind many of the AI industry’s most advanced processors. Its dominance has left rivals like Intel struggling to catch up, with meaningful market share gains appearing more like a pipe dream than measurable reality.

But in a surprising twist, Tesla CEO Elon Musk recently highlighted a big break for one of those rivals, Samsung Electronics (SSNL.F 9.01%), giving its investors some much-needed optimism. The announcement raises an important question: Will Samsung’s latest win usher in a new era of growth and pose a serious challenge to TSMC’s supremacy?

Why Samsung’s deal with Tesla matters

In late July, Musk announced on X that Tesla had signed a $16.5 billion agreement with Samsung to produce its next-generation inference chip, known as the AI6. Samsung will be manufacturing these chips at a new foundry in Texas, strategically positioning the company closer to Tesla’s headquarters and reinforcing its footprint beyond South Korea.

Tesla’s upcoming innovations — most notably its Robotaxi platform and Optimus humanoid robot — will demand highly sophisticated chip designs and huge computing capacity to function. This makes securing advanced foundry services essential for the company’s ambitions in a rapidly evolving AI landscape.

Image source: Getty Images.

How does Samsung’s relationship with Tesla impact TSMC?

At first glance, a deal of this magnitude might look like a major setback for TSMC. The reality, however, is more nuanced.

Musk clarified that TSMC will manufacture the predecessor chip to the AI6 — aptly called the AI5. In other words, Tesla is deliberately engaging with multiple foundry partners as a strategic, cautious hedge aimed at reducing supply chain risk and ensuring redundancy.

While Samsung’s win provides a boost of credibility to its lagging foundry business, analysts at Morgan Stanley said that the deal is unlikely to meaningfully dent TSMC’s dominance or serve as a material headwind to its long-term revenue and earnings potential.

Moreover, as TSMC continues to invest in its own infrastructure here in the U.S., the company remains on secure footing to deepen its ties with AI’s biggest spenders even further.

Has Samsung delivered a checkmate against its fiercest rival?

Samsung investors have gained tangible proof that strengthens the company’s long-term prospects, but TSMC’s durable technological position remains supported by entrenched scale, advanced processor leadership, and deep customer relationships. For now, this deal underscores that Samsung can still compete for landmark contracts and carve out relevance in an industry where TSMC’s gold-standard reputation remains firmly intact.

At a more macro level, the deal also signals that as AI applications become increasingly more sophisticated, leading enterprises like Tesla are keen on maintaining choice by diversifying key manufacturing partners to ensure stability, flexibility, and supply chain resilience.

For investors, the larger takeaway is clear: Samsung’s relationship with Tesla illustrates that the company is capable of winning meaningful battles. Nevertheless, TSMC is still ahead.

Rather than a checkmate, this development looks more like a fleeting stalemate at best — a dynamic that will continue to evolve as global demand for next-generation chip architectures accelerates and further intensifies the foundry race.

The semiconductor industry is the beating heart of the artificial intelligence (AI) revolution. Developing AI models wouldn’t be possible without powerful chips and advanced networking equipment, and for them to continue getting “smarter,” semiconductor suppliers will have to deliver more and more computing capacity.

For that reason, Nvidia (NVDA -0.82%) CEO Jensen Huang expects data center operators to spend up to $4 trillion on upgrading their infrastructure to meet demand from AI developers by 2030. Nvidia will be a major beneficiary of that spending over the next five years, but so will many of its peers and competitors.

Here are five semiconductor stocks to buy right now.

Image source: Getty Images.

1. Nvidia

Let’s start with the most obvious pick. Nvidia’s graphics processing units (GPUs) for data centers are the gold standard for AI development. The company just started shipping a new GPU called the GB300, which is based on its Blackwell Ultra architecture, and it’s up to 50 times more powerful in certain configurations than its flagship H100 chip, which dominated the market in 2023 and most of 2024.

The latest AI reasoning models consume significantly more tokens (words and symbols) than older one-shot large language models (LLMs), because they spend more time “thinking” in the background to weed out errors before generating outputs. This calls for more computing power, which is expected to drive explosive demand for the GB300 from the best AI developers like OpenAI, Anthropic, Meta Platforms, and xAI.

Nvidia generated a record $41.1 billion in data center revenue during its fiscal 2026’s second quarter (ended July 27), which was up 56% year over year. That number also grew by a staggering 1,081% compared to the same quarter in fiscal 2023, which was right before the AI revolution started gathering momentum. If AI infrastructure spending really does hit $4 trillion over the next five years, Nvidia will probably be one of the best stocks investors can own.

2. Broadcom

Broadcom (AVGO 0.15%) supplies AI accelerators (a type of data center chip) to at least three hyperscalers, including Alphabet. These chips have become a popular alternative to GPUs because they can be customized to suit the needs of each customer, so they offer more flexibility.

Broadcom is also a top supplier of networking equipment. Its Ethernet switches regulate how fast data travels between chips and devices, and its new Tomahawk Ultra variant delivers industry-leading low latency and high throughput, which facilitates faster processing speeds with less data loss.

Broadcom’s AI semiconductor revenue soared by 63% to $5.2 billion during its most recent quarter, but it might just be getting warmed up. The company says its three hyperscale customers plan to deploy over 1 million AI accelerators each in 2027, creating a $90 billion opportunity. Separately, a new mystery customer recently placed a $10 billion order for accelerators, and Wall Street is speculating it could be OpenAI.

3. Advanced Micro Devices

Advanced Micro Devices (AMD -0.03%) supplies chips for some of the world’s most popular consumer electronics, from Sony‘s PlayStation 5, to the infotainment systems inside Tesla‘s electric vehicles. However, the company is now laser-focused on catching up to chipmakers like Nvidia in the AI data center business.

AMD’s latest MI350 series of GPUs are based on a new architecture called Compute DNA 4, and they are 35 times faster than its previous generation that launched less than two years ago. Next year, AMD will start shipping the MI400 series, which will be paired with specialized hardware and software systems to create a fully integrated data center rack called Helios, delivering a tenfold improvement in performance relative to the MI350 series.

This highlights how quickly AMD is progressing from a technological perspective. The company is slowly capturing market share already, but these new chips could cement its position as a real player in the data center space for the long term.

4. Micron Technology

GPUs wouldn’t be as efficient without high-bandwidth memory (HBM), which stores data in a ready state to accelerate processing speeds. Simply put, more HBM capacity allows the GPU to unleash its maximum performance, which is essential in data-intensive AI workloads.

Micron Technology‘s (MU -2.83%) HBM3E solution for the data center offers industry-leading capacity and energy efficiency, and it’s embedded in Nvidia’s Blackwell Ultra GPUs and also AMD’s MI350 series. But the company will raise the bar again next year with its HBM4 solution, which will offer 60% more performance and 20% less power consumption.

Simply put, investors who believe Nvidia and AMD will sell truckloads of data center GPUs over the next five years should also be bullish on Micron’s business.

But it gets better, because some smaller AI workloads are slowly migrating to personal computers and smartphones, so they also require higher memory capacities. That’s great news for Micron because it’s a major player in those markets, too.

5. Taiwan Semiconductor Manufacturing

Finally, Taiwan Semiconductor Manufacturing (TSM -0.58%) could be the ultimate picks-and-shovels play as AI infrastructure spending ramps up. It’s the world’s largest semiconductor fabricator, and Nvidia, Broadcom, and AMD are just a few of its top clients.

Taiwan Semi offers unmatched expertise when it comes to manufacturing the most advanced chips. It works with the smallest nodes in the industry, so it can pack more transistors into each chip which is the key to unlocking processing power and energy efficiency. That is an ideal combination when it comes to AI GPUs.

Investors who own Taiwan Semi stock won’t be too concerned about which chip giant wins the AI race, because whether it’s Nvidia, Broadcom, or AMD, the demand for manufacturing capacity is only heading in one direction: up.

Published on
24/09/2025 – 9:33 GMT+2

ADVERTISEMENT

Shares in Alibaba rose around 9% in Hong Kong on Wednesday afternoon after CEO Eddie Wu said that he would lift the firm’s AI budget.

The e-commerce giant had already pledged to invest 380 billion yuan (€45bn) in AI-related infrastructure over the next three years, seeking to stay ahead as firms race to develop new models. Wu did not give details on the additional expenditure.

The pledge came as Wu was launching Alibaba’s most powerful AI model during a company conference in Hangzhou, China. The firm’s chief technology officer, Zhou Jingren, said that the Qwen3-Max model contains more than 1 trillion parameters. These are learnt values that determine how the system processes information and makes predictions.

In certain metrics, Alibaba claimed that its Qwen3-Max model outperformed rival offerings like Anthropic’s Claude and DeepSeek-V3.1, citing third-party benchmarks.

“The industry’s development speed far exceeded what we expected, and the industry’s demand for AI infrastructure also far exceeded our anticipation,” Wu said on Wednesday. “We are actively proceeding with the 380 billion investment in AI infrastructure, and plan to add more.”

Stressing that Alibaba must push ahead, Wu estimated that total global investment in AI will exceed $4 trillion (€3.4tn) in the next five years. Chinese rivals such as Tencent and JD.com, as well as US tech firms, have invested heavily in AI over the past year.

Complicating Alibaba’s progress, however, are access restrictions on AI processors from Nvidia.

Last week, China’s internet regulator banned the country’s biggest tech firms from buying Nvidia’s artificial intelligence chips, according to the Financial Times.

The reported ban comes as China seeks to boost its homegrown chip industry and wean itself off dependence on the US.

In August, Chinese firms had previously been advised not to buy Nvidia’s H20, a chip designed specifically for China, with officials in Beijing warning of perceived security risks to national data and systems.

The warning arrived after the US lifted its own ban on the export of H20 chips to China, imposed in April amid a trade spat.

In April, Navitas Semiconductor‘s (NVTS 1.48%) stock sank to an all-time low of $1.52 per share. That marked a 92% drop from its all-time high of $20.16 in November 2021. The chipmaker’s stock plummeted as it broadly missed its own long-term forecasts.

Before Navitas went public by merging with a special purpose acquisition company (SPAC) in October 2021, it claimed its revenue would surge from $12 million in 2020 to $308 million in 2024. But in 2024, the company only generated $83 million in revenue.

Image source: Getty Images.

Yet Navitas’ stock now trades at about $6. It soared over the past five months as it secured a new data center deal with Nvidia, but are those gains sustainable? Let’s review what Navitas does, why the Nvidia deal lit a fire under its stock, and if it’s worth buying.

What does Navitas produce?

Navitas produces gallium nitride (GaN) and silicon carbide (SiC) power chips, which are faster, more power-efficient, and more resistant to higher temperatures and voltages than traditional silicon chip devices. That makes them well-suited for electric vehicle (EV) chargers, data center power supplies, solar inverters, industrial motor drives, and mobile chargers. Unlike Wolfspeed, which is grappling with soaring costs as it manufactures its own SiC and GaN chips at its first-party foundries, Navitas is a fabless chipmaker that outsources its production to third-party foundries.

Navitas generates most of its revenue from its GaNFast Power ICs, which bundle together switching, sensing, control, and security features on a single chip. But in 2022, it significantly increased its exposure to the SiC market with its acquisition of GeneSiC, which develops SiC chips for the EV and data center markets.

Navitas’ top customers include PC makers like Dell and Lenovo, smartphone leaders like Samsung and Xiaomi, and Chinese EV makers like BYD and Changan. This May, Nvidia partnered with Navitas to develop more power-efficient delivery systems for its next-gen artificial intelligence data centers.

Why is Navitas’ growth cooling off?

Navitas’ sales surged in 2022 and 2023 as the GaN and SiC markets heated up, but that growth spurt ended in 2024 as it dissolved a partnership with a key distributor. Its revenue continued to decline in the first half of 2025 as its mobile and consumer markets faced seasonal headwinds and its EV, solar, and industrial customers reined in their orders to resize their inventories. Its sales in China, which accounted for 60% of its top line in 2024, are also exposed to unpredictable tariffs.

Data source: Navitas Semiconductor. YOY = Year-over-year.

Navitas’ deal with Nvidia also won’t boost its near-term revenues. It expects to ship the first samples for that collaboration in the fourth quarter of 2025, the final selections to be made in 2026, and the actual mass production of those selected chips to start in 2027.

What’s next for Navitas?

For 2025, analysts expect Navitas’ revenue to decline 42% to $48.6 million as its net loss widens to $116.4 million. For 2026, they expect its revenue to rise 9% to $53.1 million as it narrows its net loss to $78 million. That recovery should be driven by milder macro headwinds for the EV, solar, and industrial markets. But in 2027, they expect its revenue to surge 79% to $95 million as it starts mass-producing its first chips for Nvidia, while its net loss could narrow to $68 million.

However, that forecast could be too bullish because it hasn’t even shipped its first power chip samples to Nvidia. If those chips face delays or production issues, it will struggle to meet Wall Street’s rosy expectations for 2027 and beyond. But too much optimism has also been baked into its current valuations. With a market cap of $1.2 billion, Navitas is valued at 24 times this year’s sales. That price-to-sales ratio was inflated by its headline-grabbing partnership with Nvidia, but it can’t be justified by its near-term growth.

Therefore, Navitas’ stock could be cut in half and still be expensive relative to its peers. It had a great run over the past four months, but I doubt it can maintain that momentum in this wobbly market. If you don’t already own Navitas’ stock, you should probably wait for some clearer updates regarding its future roadmap before buying.

The artificial intelligence (AI) boom has been fueled by large tech companies developing impressive AI models that can handle increasingly complex tasks. But a sometimes overlooked aspect of AI are the companies that manufacture complex processors that make those models possible.

Two such semiconductor manufacturing companies are ASML (ASML -2.78%) and Taiwan Semiconductor Manufacturing (TSM -3.05%), often referred to as TSMC. While both have their strengths, which one looks like the better stock right now? Here’s what’s happening with each, and which one is likely the better AI stock.

Image source: Getty Images.

ASML’s opportunities and risks

ASML has a unique angle in the processor manufacturing market through its extreme ultraviolet (EUV) lithography system that’s used to make AI processors. These machines are very complex and not easily replicated, which is why ASML is one of the few companies in the world with these machines. This means that any semiconductor manufacturing company that needs one of these machines has to come to ASML for it.

Despite this opportunity, it’s not all sunshine and rainbows for ASML’s business. The company is reeling from President Donald Trump’s tariffs, and management said recently that potential growth in 2026 will be affected by them. ASML CEO Christophe Fouquet said on the Q2 earnings call: “We continue to see increasing uncertainty driven by macroeconomic and geopolitical developments. Therefore, while we still prepare for growth in 2026, we cannot confirm it at this stage.”

That’s a shift from management’s previous stance that the company would grow significantly this year and next. The company also lowered its estimated sales for this year to about 32.5 billion euros, down from its previous estimate of up to 35 billion euros.

That uncertainty has caused ASML’s shares to plunge recently, dropping 13% over the past 12 months. And with investors still unsure how tariffs will impact the company over the next couple of years, they’re right to be a little wary.

TSMC’s advantages and challenges

Taiwan Semiconductor also has a unique position in the AI space. The company is the leading manufacturer of AI processors, with an estimated 90% of the advanced processor market. This means that when AI giants, including Nvidia, need AI processors made, Taiwan Semiconductor is often their first choice.

This demand continues to fuel growth for the company, and TSMC’s management estimates that AI sales will double this year. The company is already well on its way, with revenue rising by 38% to $30 billion in Q2. TSMC’s bottom line is impressive as well, with earnings rising 61% to $2.47 per American depository receipt (ADR).

And while ASML is experiencing some turbulence with its business, TSMC is still going strong. Taiwan Semiconductor CEO Wendell Huang said, “Moving into third quarter 2025, we expect our business to be supported by strong demand for our leading-edge process technologies.”

Continued demand for AI processors has resulted in TSMC’s share price climbing about 40% over the past 12 months, which is significantly better than the S&P 500‘s gains of 15% over the same time. While some investors are concerned about when the AI boom will be over, it’s certainly too early to call it now.

The verdict: Taiwan Semiconductor is the better AI stock

Taiwan Semiconductor is increasing sales and earnings at a healthy clip, has a corner on AI processor manufacturing, and continues to benefit from an expanding AI market. While ASML is a strong contender, the company’s recent tariff uncertainty and lowered sales expectations aren’t great news for investors.

ASML stock is also slightly more expensive than TSMC’s at the moment, with a price-to-earnings (P/E) ratio of about 28, compared to Taiwan Semiconductor’s 26. I think both companies could be good long-term AI investments, but for all the reasons above, I think Taiwan Semiconductor deserves the win in this matchup.

Specialty chipmaker Navitas Semiconductor (NVTS -0.88%) was looking anything but special over the past few trading days. The company was the subject of a recommendation downgrade, which pushed the stock well down in price and kept it there. As of Thursday evening, Navitas’s shares were down by over 10% week to date, according to data compiled by S&P Global Market Intelligence.

Cut down to size

Although the downgrading party wasn’t a large, famous financial institution, the move nevertheless impacted Navitas stock, and not in a pleasant way. It was made on Wednesday by CJS Securities’s Jonathan Tanwanteng, who reset his recommendation on the stock to market perform — hold, in other words — from his previous ranking of market outperform (buy). He did not set a price target.

Image source: Getty Images.

Tanwanteng’s reasoning behind the downgrade wasn’t immediately apparent, but it was likely influenced by the dispiriting second-quarter results Navitas announced near the start of August.

For the period, management reported that the company suffered a year-over-year revenue decline of nearly 30%. In what was hardly more encouraging news, the company’s $0.25 per share net loss was double the deficit in the second quarter of 2024.

Memories of a hot deal fading

The resulting investor sell-off was quite the comedown for the company, which, as recently as May, was riding high on news of a deal with chip giant Nvidia. The two announced they were teaming up to develop hardware solutions for the coming wave of data centers outfitted to service the needs of artificial intelligence (AI) technology.

United States President Donald Trump has threatened to impose tariffs of up to 300 percent on semiconductor imports, with exemptions for foreign companies that commit to manufacturing in the US.

Trump has cast the proposed tariff as a way to drive investment to the US, but experts say it could also disrupt global supply chains and even penalise companies already making chips in the US.

What are the details of Trump’s plan?

Few details have been released since Trump announced plans for a 100 percent tariff at a White House event on August 7.

The US president said exemptions would be given to companies that build research or manufacturing facilities in the US, but tariffs could be applied retroactively if they failed to follow through on their planned investments.

“If, for some reason, you say you’re building, and you don’t build, then we go back, and we add it up, it accumulates, and we charge you at a later date, you have to pay, and that’s a guarantee,” Trump told reporters.

On Friday, Trump told reporters on board Air Force One that more details would be announced soon and that the tariff could be much higher than previously suggested.

“I’ll be setting tariffs next week and the week after, on steel and on, I would say chips – chips and semiconductors, we’ll be setting sometime next week, week after,” Trump said en route to Alaska to meet with Russian President Vladimir Putin.

“I’m going to have a rate that is going to be 200 percent, 300 percent,” he added.

Why does Trump want to impose tariffs on chip imports?

Trump wants to impose a tariff on chips for several reasons, but the main one is to re-shore investment and manufacturing to the US, said G Dan Hutcheson, the vice chair of Canada’s TechInsights.

“The primary goal is to reverse the cost disadvantage of manufacturing in the US and turn it into an advantage. It’s mainly focused on companies that are not investing in the US,” Hutcheson told Al Jazeera.

“Exclusions are negotiable for entities that align with his goal of bringing manufacturing back to the US.”

More broadly, the tariff is also intended to address the US dependence on imported semiconductors and buttress Washington’s position in its ongoing rivalry with China, another chip-making powerhouse.

Both issues are bipartisan concerns in the US.

The Trump administration earlier this year launched a Section 301 investigation into alleged unfair trade practices in China’s semiconductor industry, and a Section 232 investigation into the national security implications of US reliance on chip imports and finished products that use foreign chips.

Who will be impacted by the tariff?

Foreign tech giants that have already invested in the US, including the Taiwan Semiconductor Manufacturing Company (TSMC) and South Korea’s Samsung, would likely not be affected by the tariff.

It is less clear how the measure could affect other companies, including chip makers in China, where companies face barriers to US investment from both US and Chinese regulators.

Yongwook Ryu, an assistant professor at the Lee Kuan Yew School of Public Policy in Singapore, said the tariff could be used as leverage by the US as it negotiates the rate of its so-called “reciprocal tariffs” on China.

The US has imposed blanket tariffs of 10-40 percent on most trade partners since August 7, but negotiators are still hammering out a comprehensive trade deal with Beijing.

“My view is that while the reciprocal tariffs are generally aimed more at addressing the US trade deficit problem and re-shoring manufacturing back to the US, product-specific or sectoral tariffs [like semiconductors] are aimed at serving the strategic goal of strengthening US technological hegemony and containing China,” Ryu told Al Jazeera.

What is the value of US chip imports each year?

The US imported about $40bn in chips in 2024, according to a report by the American Enterprise Institute, citing United Nations trade data.

Imports mainly came from Taiwan, Malaysia, Israel, South Korea, Ireland, Vietnam, Costa Rica, Mexico and China, but experts say this data does not capture the full picture of chip flows in and out of the US.

Chips can cross borders multiple times as they are manufactured, packaged, or added to finished goods.

Chris Miller, the author of Chip War: The Fight for the World’s Most Critical Technology, estimates that another $50bn worth of chips entered the US in 2024 via products like smartphones, auto parts and home appliances from countries like China and Vietnam.

Miller also estimates that a “substantial portion” of US chip imports are manufactured in the US before being sent overseas for packaging – a labour-intensive process – and then re-imported.

“Many of the chips imported from key trading partners like Mexico, Malaysia and Costa Rica are likely actually manufactured by US firms like Texas Instruments and Intel, which have manufacturing in the US but often have their test and assembly facilities abroad,” Miller told Al Jazeera.

Why is the tariff a concern for the global chip industry?

Trump’s tariff plans have injected further uncertainty into an industry already grappling with his administration’s sweeping efforts to reorder global trade.

“It’s unclear whether the US government has the capacity to effectively enforce this and… there’s not really any guidance in terms of what these tariffs are actually going to look like,” Nick Marro, the lead analyst for global trade at the Economist Intelligence Unit, told Al Jazeera.

The White House has yet to provide details on whether the tariff will apply to chips originally made in the US and chips contained in finished products.

If the latter were included in the tariff plans, the fallout would extend to industries like electronics, home appliances, automobiles and auto parts. 

Miller said that it would be consumers in the US and elsewhere who would be among those most affected by the tariff. 

“Initially, it appears that most costs would be paid by companies via lower profit margins, though in the long run, consumers will pay the majority of the cost,” he said.

Aug. 15 (UPI) — Applied Materials’ stock price slumped by double digits on Friday after the semiconductor equipment maker reported a projected decline in revenue amid tariff worries in China.

On the Standard and Poor’s 500 index, the company’s stock decreased 11% at the opening bell and was trading at $162.09, down 13.87%, at 2 p.m. Entering trading, Applied Materials was up more than 15% for the year. The stock reached $199.29 on July 15 with the record $235.99 in April 2024.

The company, based in Santa Clara, Calif., reported the sixth consecutive quarter of revenue growth, including $7.3 billion in the third quarter, but foresees a weaker situation in the next quarter. They initially projected $6.7 billion in revenue for the quarter.

“We are expecting a decline in revenue in the fourth quarter driven by both digestion of capacity in China and non-linear demand from leading-edge customers given market concentration and fab timing,” Brice Hill, senior vice president and CFO at Applied Materials, said. “We are navigating and adapting to the near-term uncertainties by leveraging our robust supply chain, global manufacturing footprint and deep customer relationships.”

CEO Gary Dickerson, during an earnings call with analysts, said the current macroeconomic situation and trade issues have fueled “increasing uncertainty and lower visibility,” mainly within its business in China.

In addition, he said their forecast does not account for pending export license applications and a substantial backlog of products.

Dickerson noted the easing of spending from customers, with Chinese clients cutting spending after increasing equipment manufacturing in the region.

President Donald Trump has proposed a 100% tariff on semiconductors and possibly a 300% rate. Exempt companies would be those with manufacturing facilities in the United States.

Applied Materials doesn’t make chips, and instead supplies equipment, services and software used by the makers of the chips. The company’s largest plant for logistics and logistics is in Austin, Texas.

On Monday, Trump extended a tariff pause until Nov. 10 on products sent to the United States from China. Originally, he threatened 145% duty, but it was later lowered to 30% plus the baseline tariffs imposed on nearly all U.S. trading partners. The baseline remains in effect.

In June, Trump announced a trade agreement with China over rare earth minerals. Under the deal, China would export rare earth minerals to the United States with both countries reducing their tariffs for 90 days. Rare earth minerals fuel energy sources for mobile devices and electric vehicles.

Despite uncertainty, Applied Equipment in its report wrote that “we remain very confident in the longer-term growth opportunities for the semiconductor industry and Applied Materials.

The company’s adjusted earnings of $2.11 per quarter was short of the $2.39 expected by LSEG.

Net income hit $1.78 billion, or $2.22 per share. One year ago, it was $1.71 billion, or $2.05 per share.

The gross margin was 48.8% compared with 47.3% one year ago, and the operating margin was 30.6% vs. 28.7% in 2024.

The company specializes in materials engineering solutions for semiconductors, flat panel displays and solar photovoltaic industries. The company’s revenue in semiconductor equipment is No. 1 in the world, followed by the Dutch company ASML.

Sales at all three Applied Materials units rose: Semiconductor Systems at $5.43 billion, Applied Global Services at $1.60 billion and and Display t a$263 million.

The company’s market capitalization is $151.06 billion. It was founded in 1967 as a startup.

By&nbspAP with Eleanor Butler

Published on
08/08/2025 – 9:20 GMT+2

ADVERTISEMENT

Intel shares slumped on Thursday after President Donald Trump said in a social media post that the chipmaker’s CEO needed to resign.

“The CEO of Intel is highly conflicted and must resign, immediately,” Trump posted on Truth Social. “There is no other solution to this problem. Thank you for your attention to this problem!”

Trump made the post after Senator Tom Cotton sent a letter to Intel Chairman Frank Yeary, expressing concern over CEO Lip-Bu Tan’s investments and ties to semiconductor firms that are reportedly linked to the Chinese Communist Party and the People’s Liberation Army. Cotton asked the board whether Tan had divested his interests in these companies to eliminate any conflicts of interest.

It’s not immediately clear if Tan, who took over as Intel’s CEO in March, has done so.

In a statement, Intel said it was “deeply committed to advancing US national and economic security interests”. The firm said it was making “significant investments aligned with the President’s America First agenda”.

Cotton’s allegations

“In March 2025, Intel appointed Lip-Bu Tan as its new CEO,” Cotton wrote in the letter. “Mr. Tan reportedly controls dozens of Chinese companies and has a stake in hundreds of Chinese advanced-manufacturing and chip firms. At least eight of these companies reportedly have ties to the Chinese People’s Liberation Army.”

Cotton specifically called out Tan’s recent leadership of Cadence Design Systems in the letter. According to the US Department of Justice, Cadence, agreed in July to plead guilty to resolve charges that it violated export controls rules to sell hardware and software to China’s National University of Defense Technology, which is linked to the Chinese military.

Tan was the CEO of Cadence when the company violated the rules between 2015 and 2021.

The US Department of Commerce’s Bureau of Industry and Security also fined Cadence $95 million for the same breaches, saying Cadence admitted that “employees of its Chinese subsidiary knowingly transferred sensitive US technology to entities that develop supercomputers in support of China’s military modernisation and nuclear weapons programs.”

Cadence did not immediately respond to AP requests.

The digital race

Tan previously launched the venture capital firm Walden International in 1987 to focus on funding tech start-ups, including chip makers.

China’s state media has described Tan as “actively” devoted to Chinese and Asian markets, having invested not only in the Taiwan Semiconductor Manufacturing Company, but also China’s state-owned enterprise SMIC, which seeks to advance China’s chipmaking capabilities.

The demands made by Trump and Cotton come as economic and political rivalries between the US and China increasingly focus on the competition over chips, AI and other digital technologies that experts say will shape future economies and military conflicts.

Cotton, the chairman of the Senate Intelligence Committee, has raised concerns that Chinese spies could be working at tech companies and defence contractors, using their positions to steal secrets or plant digital backdoors that give China access to classified systems and networks.

On Thursday the Arkansas Republican wrote to the Department of Defense, urging Defense Secretary Pete Hegseth to ban all non-US citizens from jobs allowing them to access DoD networks. He has also demanded an investigation into Chinese citizens working for defence contractors.

“The US government recognises that China’s cyber capabilities pose one of the most aggressive and dangerous threats to the United States, as evidenced by infiltration of our critical infrastructure, telecommunications networks, and supply chains,” Cotton wrote in an earlier letter, calling on the Pentagon to conduct the investigation.

National security officials have linked China’s government to hacking campaigns targeting prominent Americans and critical US systems.

“US companies who receive government grants should be responsible stewards of taxpayer dollars and adhere to strict security regulations,” Cotton wrote on the social platform X.

Playing catch-up

Intel had been a beneficiary of the Biden administration’s CHIPS Act, receiving more than $8 billion (€6.9bn) in federal funding to build computer chip plants around the country.

Shares of the California company slid 3.5%, while markets, particularly the tech-heavy Nasdaq, gained ground.

Founded in 1968 at the start of the PC revolution, Intel missed the technological shift to mobile computing triggered by Apple’s 2007 release of the iPhone, and it has lagged behind more nimble chipmakers. Intel’s troubles have been magnified since the advent of artificial intelligence — a booming field where the chips made by once-smaller rival Nvidia have become tech’s hottest commodity.

Intel is shedding thousands of workers and cutting expenses, including some domestic semiconductor manufacturing capabilities, as Tan tries to revive the fortunes of the struggling chipmaker.

US President Donald Trump said the tariff will not impact companies if they have already invested in US facilities.

United States President Donald Trump says he will impose a 100 percent tariff on foreign-made semiconductors, although exemptions will be made for companies that have invested in the US.

“We’ll be putting a tariff on of approximately 100 percent on chips and semiconductors, but if you’re building in the United States of America, there’s no charge, even though you’re building and you’re not producing yet,” Trump told reporters at the Oval Office on Wednesday evening.

The news came after a separate announcement that Apple would invest $600bn in the US, but it was not unexpected by US observers.

Trump told CNBC on Tuesday that he planned to unveil a new tariff on semiconductors “within the next week or so” without offering further details.

Details were also scant at the Oval Office about how and when the tariffs will go into effect, but Asia’s semiconductor powerhouses were quick to respond about the potential impact.

Taiwan, home of the world’s largest chipmaker TSMC, said that the company would be exempt from the tariff due to its existing investments in the US.

“Because Taiwan’s main exporter is TSMC, which has factories in the United States, TSMC is exempt,” National Development Council chief Liu Chin-ching told the Taiwanese legislature.

In March, TSMC – which counts Apple and Nvidia as clients – said it would increase its US investment to $165bn to expand chip making and research centres in Arizona.

South Korea was also quick to extinguish any concerns about its top chipmakers, Samsung and SK Hynix, which have also invested in facilities in Texas and Indiana.

Trade envoy Yeo Han-koo said South Korean companies would be exempt from the tariff and that Seoul already faced “favourable” tariffs after signing a trade deal with Washington earlier this year.

TSMC, Samsung and SK Hynix are just some of the foreign tech companies that have invested in the US since 2022, when then-President Joe Biden signed the bipartisan CHIPS Act offering billions of dollars in subsidies and tax credits to re-shore investment and manufacturing.

Less lucky is the Philippines, said Dan Lachica, president of Semiconductor and Electronics Industries in the Philippines Foundation.

He said the tariffs will be “devastating” because semiconductors make up 70 percent of the Philippines’ exports.

Trump’s latest round of blanket tariffs on US trade partners is due to go into effect on Thursday, but the White House has also targeted specific industries like steel, aluminium, automobiles and pharmaceuticals with separate tariffs.

Published on
16/07/2025 – 10:18 GMT+2

ADVERTISEMENT

Supplier of chipmaking equipment ASML retracted its growth forecast for the coming year on Wednesday, sending shares down around 7% in morning trading in Amsterdam.

“The level of uncertainty is increasing, mostly due to macroeconomic and geopolitical considerations. And that includes, of course, tariffs,” said CEO Christophe Fouquet. “Therefore, while we still prepare for growth in 2026, we cannot confirm it at this stage.”

The warning came despite the fact that the Dutch firm saw sales and bookings rise above analysts’ expectations during the second quarter.

Sales rose 23% to €7.7 billion, while net bookings came in at €5.5bn. Net income was at €2.3bn.

For the third-quarter, ASML predicted net sales between €7.4bn and €7.9bn, falling short of estimates, and a gross margin between 50% and 52%.

The firm also forecast 15% revenue growth for the year ahead.

A boom in artificial intelligence is fuelling demand for ASML’s semiconductor-making machines, which are needed to power AI technologies.

Last week, chipmaker Nvidia — a firm that relies on ASML products — became the first company in the world to reach a market value of $4 trillion.

So far, the extent to which ASML will be affected by US tariffs and retaliatory duties is unclear. Semiconductors are currently exempt from Trump’s duties although it’s not yet known whether chipmaking machines will receive the same leniency.

Easing tensions between the US and China are also helping Nvidia, which in turn bodes well for ASML. On Tuesday, Nvidia said it would start selling its H20 AI chip in China again after the Trump administration relaxed export restrictions. The move is a U-turn for the government, which in April banned sales of the chip to China, linked to concerns that the technology could be used for military purposes.

ASML also faces restrictions on sending certain advanced products to China. There has been no suggestion that these measures, imposed by the Dutch government, will be lifted.

“ASML cites the macroeconomic environment and tariffs having an impact on the orders. More specifically, it is more likely uncertainty from China, memory capex uncertainty and the struggles at Intel and Samsung that are more likely to be hampering things,” said Ben Barringer, global technology analyst at Quilter Cheviot.

Intel and Samsung, two ASML customers, are facing financial headwinds, with the latter reporting its first fall in profit in around two years last week.

Barringer continued: “Ultimately, this is a speed bump for what remains a high-quality company. It still has a big backlog so growth should still pull through”.

BOSTON — A federal judge has blocked the Trump administration from making drastic cuts to research funding provided by the National Science Foundation.

U.S. District Judge Indira Talwani in Boston on Friday struck down a policy change that could have stripped universities of tens of millions of dollars in research funding. The universities argued that the move threatened crucial work in artificial intelligence, cybersecurity, semiconductors and other technology fields.

Talwani said the change, announced by the NSF in May, was arbitrary, capricious and contrary to law.

An email Saturday to the National Science Foundation was not immediately returned.

At issue are “indirect” costs, expenses such as building maintenance and computer systems that aren’t linked directly to a specific project. Currently, the National Science Foundation determines each grant recipient’s indirect costs individually and is supposed to cover actual expenses.

The Trump administration has dismissed indirect expenses as “overhead” and capped them for future awards by the National Science Foundation to universities at 15% of the funding for direct research costs.

The University of California, one of the plaintiffs, estimated the change would cost it nearly $100 million a year.

Judges have blocked similar caps that the Trump administration placed on grants by the Energy Department and the National Institutes of Health.