More important than Nvidia's earnings is the growth of AI frontier companies

Nvidia's earnings release scheduled for next week is expected to test whether Nvidia can keep leading the market rally. It is seen as an important indicator of where Nvidia stands in the US-China AI competition.

Nvidia has posted earnings beats for four consecutive quarters. Yet shares fell right after three of those four releases, a paradox analysts attribute to the growing difficulty of clearing the unofficial 'whisper number' that investors expect.

Memory stocks are currently driving gains, but signs suggest memory margins are close to peaking. Historically, when margin growth slows, share prices have tended to follow lower, raising concerns about the chip sector as a whole.

When growth stocks approach a peak, companies tend to expand share buybacks. Nvidia has also been increasing its buyback size. Buybacks can support the stock in the short term, but analysts warn they may lead to declines if doubts grow about long-term growth.

This could challenge Nvidia's dominant position.

TSMC's conservative investment stance is creating a supply bottleneck in AI chips. This is making it hard for Nvidia's major customers to secure chips, and the shortage could give rivals such as Intel an opening to enter the foundry market, changing Nvidia's competitive landscape.

The key point to watch in Nvidia's earnings release is whether it can keep its high gross margin, currently around 75%. As AI infrastructure costs rise, the central question is whether Nvidia can pass those higher costs on to customers in the way Apple has, defending its margins. This will be an important gauge of Nvidia's long-term profitability.

Intel, a pioneer of the memory market, had historically built its business around processors. Its recent move into memory is being questioned as a political maneuver to placate customers rather than a sincere technological push, and it may also be read as an effort to stabilize memory prices.

Drawing on Lee's experience helping develop HBM at SK hynix, Intel aims to use its memory manufacturing know-how to raise foundry productivity.

With TSMC's production capacity stretched, the market is focused on whether Intel Foundry can emerge as an alternative. The analysis holds that technological progress in foundry could pave the way to expanded CPU production.

For decades Intel reigned over the semiconductor industry as a 'super-powerful' company wielding strong influence. There are concerns that this corporate culture could clash with the customer-oriented model that foundry business requires. Intel's past conduct, including failing to deliver on its 2023 roadmap commitments, supports such skepticism. Among global chip companies, many take a negative view of the likelihood that Intel's foundry business will succeed.

The foundry market has hardened into a structure in which TSMC's dominance makes it very hard for latecomers to catch up. Samsung Electronics' foundry market share has fallen from roughly 15% in the past to the 7% range today. This illustrates how structurally difficult it is to catch up with TSMC in foundry, and reflects the realistic situation of having to be satisfied with maintaining operations and generating modest profits.

Intel CEO Pat Gelsinger pursued an aggressive strategy to expand foundry operations across the world, including Germany and Southeast Asia. That reckless expansion, however, drew market concern and ultimately led to Gelsinger's ouster. Some assessments hold that without massive support from the US government, Intel's foundry division would already have collapsed.

It struck partnerships with major companies such as Naver and pushed hard, but it has produced no meaningful results so far and has effectively failed.

Samsung Electronics, by contrast, has underinvested in legacy processes and is struggling to chase TSMC, according to the analysis.

Some on Wall Street read the recent shareholder return policies of Korean chipmakers such as Samsung Electronics and SK hynix as a sign of limited growth. The argument is that when growth drivers are sufficient, companies should focus on technology investment rather than shareholder returns. A company starting to expand shareholder returns can suggest internal growth problems.

But buybacks alone rarely drive sharp share price gains. This suggests the effect of shareholder return policies on the stock price may be limited.

Analysts argue that Nvidia's earnings announcement next week will not be enough to lift semiconductor stocks. The real trigger for share price gains, they say, depends on the actual growth of AI frontier companies such as Anthropic and the infrastructure demand they generate. Their revenue growth and ecosystem expansion are expected to ultimately be the core driver that lifts the entire semiconductor market.