India Chip Push, Huawei vs. Nvidia, and SpaceX's Orbital Bet

India’s semiconductor ambitions moved from aspiration to capital commitments this week, Applied Materials put $5 billion behind the bet, and Huawei signaled it wants a serious fight with Nvidia on AI hardware. Meanwhile, SpaceX is counting down to a milestone test that carries real commercial stakes.

Key points

• Applied Materials announced a $5 billion plan to build a chip R&D lab and supply chain presence in India, coinciding with Tata Electronics signing seven MoUs with semiconductor supply-chain companies at the same event.
• Micron CEO Sanjay Mehrotra said the company’s Sanand, Gujarat facility is already producing “Made in India” memory chips and announced plans for a significant scale-up, calling memory critical to AI-era growth.
• Huawei used its annual Shanghai summit to accelerate its AI chip roadmap and, through Rotating Chairman Wang Tao, claim it is narrowing the software gap with Nvidia.
• Meta is expanding use of its in-house AI chips to reduce costs, a move that could trim third-party silicon spend over time.
• SpaceX has targeted September 22 for Starship’s 14th test flight, which would be its first orbital attempt and its first carrying a revenue-generating payload.

Is India becoming a credible node in the AI chip supply chain?

The concentration of announcements at SEMICON India 2026 this week is notable. Applied Materials’ $5 billion commitment covers both R&D infrastructure and supply chain buildout, not just a symbolic research center. Tata Electronics, which is developing India’s first commercial chip fabrication plant, signed seven MoUs with supply-chain partners at the same gathering, suggesting real ecosystem coordination rather than isolated announcements.

Micron’s Sanand plant adds a tangible proof point. The facility is already shipping product, and Mehrotra’s language, “from vision to execution,” tracks with Micron’s broader push to diversify assembly and testing outside China and Taiwan. For investors watching the AI memory trade, an Indian production ramp means Micron is building geographic redundancy into a segment it considers central to AI infrastructure demand.

UST COO Gilroy Mathew argued at the same conference that India can bypass legacy semiconductor models entirely and build AI-native manufacturing from the ground up. That is an optimistic read, but the capital flowing in gives it more grounding than it had a year ago.

What does Huawei’s software claim mean for Nvidia’s moat?

Nvidia’s hardware lead in AI accelerators is well understood. Its real moat, as most analysts frame it, is CUDA, the software ecosystem that makes its chips the default choice for model training and inference. Huawei’s claim at its Shanghai summit that it is narrowing that software gap is worth watching, even if it should be treated with appropriate skepticism given the source.

China’s AI developers face continued pressure to qualify domestic hardware as export controls persist. That creates a captive testing ground for Huawei’s stack. If Chinese hyperscalers and model labs can run workflows efficiently on Huawei silicon, the company gains real-world validation that no internal benchmark can replicate. The question for Nvidia investors is not whether Huawei displaces Nvidia globally in the near term (it almost certainly will not), but whether it locks in Chinese AI spend that Nvidia can no longer access. That is a ceiling on Nvidia’s addressable market, not a floor on its current business.

Meta’s separate move to expand in-house chip usage to cut costs is a reminder that even Nvidia’s largest customers are working to reduce dependence. Custom silicon at hyperscale takes years to fully displace merchant chips, but the direction of travel among major buyers is consistent.

What is actually at stake in SpaceX’s September 22 Starship test?

Starship’s 14th test flight is structurally different from the 13 that preceded it. Per Motley Fool’s analysis, this would be the rocket’s first orbital attempt and its first flight carrying a payload that generates revenue. That changes the calculus from pure engineering milestone to commercial proof-of-concept.

Separately, SpaceX has been pitching orbital AI data centers, with CEO Elon Musk targeting a late-2027 launch window according to a September 6 CNBC report. Industry analysts are more cautious, pointing to four unresolved technical problems: heat dissipation in a vacuum, radiation hardening, debris risk, and the sheer cost of getting and keeping hardware in orbit. Analysts cited in the Yahoo Finance piece put real scale for orbital data centers in the 2030s, not 2027. For investors in SPCX, the September 22 test matters less for the data center thesis than for SpaceX’s core launch business, including NASA contracts and Starlink deployment cadence.

Apple’s valuation question

A Seeking Alpha piece published today downgraded Apple while stopping short of a sell, flagging that the stock trades at roughly 36x forward earnings. The author acknowledges Apple’s pricing power and AI-driven premium product cycle but flags a reality check ahead. At 36x, Apple is priced for consistent execution with limited margin for disappointment, a tension that is not new but becomes sharper as the AI hardware upgrade cycle faces tougher year-over-year comparisons.

This is not investment advice. Investors should conduct their own due diligence before making any trading decisions.