AI Stocks & Markets

Today's Stories

1. 1

   Broadcom vs. Marvell: Custom AI Chip Partners Face Valuation Divide

   Broadcom and Marvell Technology, the two companies building custom chips for hyperscalers like Google, Meta, and Apple, pursue different strategies—Broadcom assembles a diverse customer roster and operates infrastructure software alongside chip design, while Marvell focuses on custom chips wrapped in optical interconnect technology and has acquired interconnect specialists to strengthen its position. As cloud providers race to control their own chip destiny rather than depend entirely on Nvidia, these two companies dominate the market for turning custom designs into working silicon. Broadcom brings steadier diversification (its software business cushions chip-market swings), while Marvell offers a more specialized, faster-growing bet on the custom-chip boom—but trades at a richer valuation that leaves little margin for error.

   The market prices Marvell at a much richer valuation than Broadcom despite the latter's dominance, signaling investor expectations for faster growth from the smaller company. Both firms depend heavily on a concentrated group of hyperscaler customers whose spending could shift, and Marvell's acquisitions must integrate successfully to justify its premium price.

2. 2

   AI Bubble Risks Burst as Tech Limits Emerge, Analysts Warn

   Investment advisers and financial analysts are increasingly warning that an AI investment bubble will eventually burst, with high-profile figures like Jeremy Grantham announcing plans to sell tech shares over concerns that AI deployment has outpaced realistic utility. The warning comes as users report losing confidence in certain AI services and companies discover limits to AI's capabilities in real-world applications. AI has driven investor focus toward a narrow set of companies—Amazon, Alphabet, Nvidia, Meta, Microsoft, Apple, and Tesla—on major indices like the S&P 500 and Nasdaq. If confidence erodes, the concentrated bets on these stocks could unwind sharply. For businesses, the concern is concrete: AI works well in stable, predictable tasks, but manufacturing and other sectors face challenges—late supplier deliveries, machine failures, fluctuating demand, regulatory constraints—that existing AI cannot yet address, suggesting overinvestment relative to near-term payoff.

   The disconnect between AI's perceived and actual capabilities is widening. Companies have invested heavily in integrating AI into operations, but as the technology encounters real operational complexity, the value proposition may shift from a revolutionary transformation to a narrower utility—much like electricity or railroads, where profits accrue primarily to service builders rather than the infrastructure itself.

3. 3

   Wall Street splits on chip stocks: bullish Marvell, cautious Intel

   Major Wall Street banks have taken opposite stances on two AI chip companies. Goldman Sachs raised its price target on Marvell Technology, citing improving visibility into its custom-silicon pipeline, while Morgan Stanley and Bank of America have expressed skepticism toward Intel's ability to generate attractive returns from its foundry business. Marvell has become a favored way to play AI infrastructure through two reinforcing businesses—custom silicon design partnerships with major cloud operators and optical interconnect (data-shuttling hardware for AI data centers). Marvell now expects to supply interconnect to all five of the largest U.S. cloud operators. Intel, by contrast, is attempting a more difficult turnaround, trying to fix its product lineup while building a contract manufacturing business simultaneously, and Wall Street is uncertain whether its advanced manufacturing process will turn a profit for another couple of years.

   Marvell's addition to the S&P 500 in June marked its arrival as a major player. Neither thesis is risk-free—Marvell's growth depends on a handful of enormous customers who could pull work in-house, while Intel's 18A-P process entered risk production on schedule this summer with resolved yield problems, though skeptics worry about foundry economics rather than engineering capability.

4. 4

   Microsoft's Best Buy Yet, Says Analyst—Azure Cloud Driving 40% Growth

   Microsoft's Azure cloud business grew 40% year over year in its fiscal third quarter (ended March 31, 2026), matching growth rates from the prior two quarters. The company's AI business surpassed an annual revenue run rate of $37 billion（約5.9兆円）, up 123% year over year, while commercial remaining performance obligations (contracted work not yet recognized as revenue) roughly doubled to $627 billion（約100兆円）. Microsoft's entrenchment in enterprise software—Office, Windows, Teams, developer and security tools—means businesses already depend on the company and are now adopting its AI services through the same vendor relationship. That customer stickiness, combined with heavy cash generation (fiscal Q3 net income climbed 23% to $31.8 billion（約5.1兆円）), makes Microsoft a lower-risk way to own the AI trend compared to chipmakers exposed to volatile demand swings.

   The company plans to invest about $190 billion（約30兆円） in capital expenditures in calendar year 2026, including about $25 billion（約4兆円） from higher component pricing. That massive spending, which will later hit profits as depreciation, is the main reason the stock trades about 30% below its 52-week high of $555.45—investors want proof the build-out will justify its cost.

5. 5

   Palantir on track for $400B valuation by end of 2027, analysts say

   Palantir Technologies, valued at roughly $300 billion（約48兆円） this summer, could reach approximately $400 billion（約64兆円） by the end of 2027. The company has expanded its AI Platform offerings this year, including making its AIP Analyst tool broadly available for non-technical employees to query data in plain language, and rolling out support for AI agents to autonomously build and edit applications across its Foundry platform. Palantir is shifting from being perceived as a consulting shop to embedding itself into daily business operations through its AI Platform, which customers build on rather than simply test. This stickiness—demonstrated by wins like signing Wheels Up as a launch customer for a private aviation operating system—could support higher valuations over time, though the company still depends significantly on U.S. government work and stock-based compensation that dilutes existing shareholders.

   The path to $400 billion（約64兆円） depends on commercial growth offsetting structural risks including reliance on U.S. government budgets, shifting political priorities, and the need to move beyond government revenue. Even after its pullback, Palantir trades at a valuation that assumes years of rapid growth, leaving little room for a soft quarter, delayed contract, or broader cooling in AI spending.

6. 6

   Meta faces lawsuit over addictive platform design; Cantor Fitzgerald maintains Buy rating

   On June 30th, a California judge rejected Meta Platforms Inc.'s application to dismiss a lawsuit filed by state attorneys accusing the company of designing platforms to be addictive to teenagers. The company is also reported to be selling excess computing capacity amid ongoing debate about its AI spending. Meta's stock is down 13% over the past year and 2.9% year-to-date, and the addiction lawsuit adds legal risk to the company's operations. At the same time, Cantor Fitzgerald reiterated a Buy rating with a $750 share price target, signaling confidence in the company's artificial intelligence strategy despite reported concerns from its CEO about the pace of progress in agentic AI (systems that make decisions and complete tasks autonomously).

   Cantor Fitzgerald's $750 share price target and the ongoing lawsuit outcome, which could affect both the company's reputation and regulatory environment.