Nvidia faces dual headlines: SpaceX's exclusive selection of Nvidia's Vera Rubin chips signals massive demand, but memory supply doubts linger.
The market has a peculiar way of picking sides. When Nvidia's stock climbed 2.18 percent to 193.96 euros on a day that brought both a landmark SpaceX contract and fresh reports of memory-chip compromises, investors effectively voted with their wallets: the satellite deal carries more weight than supply-chain concerns.
That conviction is visible in the broader tape. The shares have gained 11.39 percent over the past week, trading 8.18 percent above the 50-day moving average of 179.29 euros—a technical signal that the market is accepting Nvidia's narrative rather than the skeptics'. The RSI reading of 63.7 suggests the stock is approaching overbought territory without having crossed into it.
The headline development is Elon Musk's decision to make Nvidia the exclusive compute partner for his satellite ambitions. Starting in 2027, SpaceX will deploy Nvidia's Vera Rubin architecture—pairing Rubin GPUs with Vera CPUs—aboard its Starmind-AI1 satellites. The choice is striking because SpaceX had evaluated alternative hardware options before settling on a single-vendor approach.
For Nvidia bulls, this represents the strongest evidence yet that the CUDA ecosystem remains a moat that AMD and others cannot easily cross. The decision also arrives as investors are rotating decisively toward high-margin AI names with integrated business models, a shift that has left memory makers like Western Digital nursing post-earnings losses.
Yet the same week brought uncomfortable reporting from SemiAnalysis and The Information. Nvidia may need to slash the memory configuration on its upcoming Rubin Ultra chip from a planned one terabyte of HBM4E to as little as 192 gigabytes in some variants. The culprit: HBM4E components are scarce and expensive, reportedly accounting for more than half of total material costs.
Nvidia's response has been to frame the changes as "continuous optimization" rather than a retreat, insisting the roadmap remains intact. That language invites skepticism, but there is a reasonable case for taking the company at its word. Managing a supply constraint openly is not the same as losing an innovation edge, and the stock's position 4.22 percent below its 52-week high of 202.50 euros suggests the market has not priced in a meaningful derailment.
The memory concerns must be weighed against the sheer scale of Nvidia's recent dealmaking. On Monday, the company and South Korea's SK Group unveiled a partnership valued at more than $500 billion to build AI infrastructure for data-center demand. SK Telecom will construct a two-gigawatt "AI factory" equipped with Nvidia's Vera Rubin platform, while SK hynix will collaborate on next-generation HBM chips. The first facility is slated to come online in 2027.
That announcement followed a July 29 deal with Safe Superintelligence. CEO Jensen Huang has been characteristically bullish, declaring that Vera Rubin chips are "already in production" with "massive volumes" to follow. The cadence of announcements suggests an industry that feels compelled to escalate its superlatives on a weekly basis.
Not everyone is convinced. On July 31, reports emerged that Nvidia is in talks to backstop $250 billion in OpenAI's data-center leasing commitments—reviving concerns about circular financing, where Nvidia effectively funds its own customers so they can buy its chips. The stock fell roughly five percent that day, dragging the broader chip sector down and ceding the title of world's most valuable company to Apple.
A persistent China problem adds to the headwinds. Export controls have reduced Nvidia's China revenue contribution in data-center business to effectively zero, according to a late-June report—a market that once contributed billions has vanished, offset only by the sheer scale of U.S. and Asian orders.
Insider activity presents a quieter question mark. Media reports indicate executives have sold roughly $410 million worth of shares over the past three months. Diversification is a legitimate motive, but combined with financing-structure doubts, it forms a second, softer note of caution. Star investor Michael Burry amplified that theme on Tuesday, flagging a potential $250 billion demand risk—the same figure at the center of the OpenAI negotiations.
The stock has recovered substantially from its late-July stumble. Thursday's close of 189.82 euros sits roughly nine percent above the prior week's level, though still about six percent below the 52-week high. Institutional conviction remains a stabilizing force: Tiger Global Management increased its Nvidia stake by 9.1 percent to roughly $2.09 billion in the first quarter of 2026, with Lazard Asset Management following suit.
The analyst community is similarly one-sided: 36 analysts currently rate the stock a buy, with an average price target of 261.99 euros, implying 35.1 percent upside. Nvidia's market capitalization stands at 4,602.95 billion euros.
The real test arrives on August 26, when Nvidia reports fiscal second-quarter results for the period ending July 26. The market expects revenue near $91 billion. The question is no longer whether Nvidia can deliver—it's whether the customers financing all these data centers can do the same.