New York, New York | July 9, 2026
Nvidia shares are hovering near their lowest point of the year, and the stock that once led every rally is now taking the brunt of market criticism. This contradiction is at the center of the latest Jim Cramer Nvidia stock July 2026commentary, delivered this week on CNBC’s “Squawk on the Street.” Cramer’s blunt verdict cuts against the prevailing mood on trading desks: Cramer said everything revolves Nvidia, he told viewers, even as the chipmaker limps through one of its worst stretches of the year. The remark landed as Nvidia shares dropped about 2% for the day, leaving the stock only slightly up for 2026 and trading at just under 19 times forward earnings, its lowest valuation since January.
It is an odd moment for the company that built the AI trade. Nvidia has become, in Cramer’s words, the stock “everyone hates,” yet he insists the selling has run well ahead of the fundamentals. The Nvidia stock-lagging-2026 narrative has taken hold across trading floors for a specific reason: Nvidia’s own customers keep publicly complaining about what they pay for its chips while quietly building their own silicon.
Why the Chip Giant Still Sets the Terms
Cramer’s point is often overlooked in the daily rush of market news. Nvidia’s biggest customers, including Google, Amazon, and Tesla, have all complained about prices, but they still rely on Nvidia’s technology for their data centers. Alphabet and Amazon say their own chips can compete with Nvidia’s, but Cramer is openly skeptical of these claims. Broadcom, led by CEO Hock Tan, is seen as a real competitor that could take some market share. Still, according to Cramer, none of these moves have pushed Nvidia out of its central role in building AI infrastructure. Reuters also reported that China’s DeepSeek is now working on its own AI chips, which adds to the negative outlook. But Cramer keeps saying that doubting Nvidia’s importance is not the same as proving it’s no longer important.
Some of the recent pressure on Nvidia stems from a SemiAnalysis report claiming manufacturing delays for Nvidia’s next-generation Kyber server system. The report suggests the launch could be pushed from 2027 to 2028. Nvidia directly denied this, saying its roadmap “remains intact.” Cramer saw this denial as a strong reason to buy the stock on the dip and told investors so on social media. He also offered another explanation for Nvidia’s weak performance: the stock might be a “source of funds,” meaning investors are selling it to raise cash for other opportunities, like the newly public SpaceX before it joins the Nasdaq 100.
The Trade Has Rotated, Not Disappeared
What makes this moment distinct is where the money has actually gone. On a recent Cramer Mad Money Nvidia AI infrastructure segment, the host laid out a framework that reframes the entire AI trade rather than simply defending one stock. Wall Street, he argued, is now rewarding the companies that supply the picks and shovels of artificial intelligence while punishing the hyperscalers footing the bill for that infrastructure. That is the essence of Wall Street rewarding AI suppliers not customers: Micron, Marvell, AMD, SanDisk, and Intel have posted some of the sector’s strongest gains this year, even as the so-called Magnificent Seven have shed roughly $2.3 trillion in combined market value over a single month. Micron’s gross margin has expanded from 39% to nearly 85% year over year as memory prices have, in Cramer’s phrase, gone through the roof. SanDisk has rallied more than eightfold year to date amid surging data center demand. Cramer named Intel his current favorite stock, citing CEO Lip-Bu Tan’s turnaround and disclosing that his Charitable Trust initiated a position in June, adding to it twice since.
Nvidia still aligns with Cramer’s idea of a supplier that benefits from selling products in short supply and high demand. He argues that Nvidia’s weak performance this year is more about competition concerns than any real problems with its core business, and the numbers support his view.
What the Next Product Cycle Actually Costs
The strongest proof that Nvidia is still at the center of AI spending is found in its bill of materials, not its stock price. According to Morgan Stanley, the Nvidia VR200 NVL72 rack $7.8 million estimate for the Vera Rubin platform is estimated to cost $7.8 million, almost double the $4 million that hyperscalers paid for the previous GB300 generation. This increase isn’t mainly due to GPUs, though their prices have also risen Nvidia is expected to charge about $55,000 per Rubin GPU and around $5,000 per Vera CPU when sold in bulk to large buyers.
Memory is where the real inflation shows up. Morgan Stanley estimates memory content per rack has surged to around $2 million, pushing memory’s share of the total bill of materials to roughly 25%, up from just 5% to 10% in the prior generation. That single data point explains why Micron and SanDisk have outperformed Nvidia this year: as the GPU’s share of total system cost falls from roughly 63% to 51%, memory suppliers gain a larger slice of every dollar hyperscalers spend building out AI capacity. First shipments of the new rack are scheduled for the third quarter of 2026, with volume production ramping in the fourth quarter, meaning the cost structure Cramer is describing will begin showing up in hyperscaler capital budgets within months, not years.
Cramer’s Other Conviction Call: Power, Not Just Chips
Cramer’s view goes beyond just semiconductors. On a recent Mad Money lightning round, a caller asked about GE Vernova, and Cramer did not hesitate. The Nvidia GE Vernova Cramer buy connection reflects his wider thesis that AI infrastructure spending flows well beyond chipmakers into the power and electrification companies that keep data centers running. Cramer said his Charitable Trust has “a very big position” in GE Vernova and repeated his buy rating, even though the stock is trading near record highs above $1,100.In the first quarter of 2026, GE Vernova reported $9.3 billion in revenue, up 16% from last year, with orders rising 71% to $18.3 billion and a backlog of $163 billion. The company’s management has since raised its full-year outlook, citing growing demand from both hyperscalers and utilities for reliable power.
Taken together, Cramer’s commentary this week amounts to a Jim Cramer “everything revolves around Nvidia despite lagging stock” July 9 2026 analysis that reaches well beyond a single ticker. His more general claim is that Cramer Nvidia still central AI trade Wall Street rewarding suppliers not customers 2026 describes an entire market structure, one in which value has migrated toward the companies selling scarce components memory, networking silicon, and power generation rather than the household names spending hundreds of billions to build AI data centers.
The Road Ahead
None of this settles the short-term uncertainty around Nvidia’s stock. The company’s next earnings report will be the real test of whether manufacturing issues are impacting data center revenue and if the Kyber delay claims are more than just one report. What’s clearer is that demand for AI infrastructure stays strong. Hyperscalers are not slowing down their AI projects; they’re just shifting their spending more toward memory, power, and networking than they did last year. For investors comparing Cramer’s confidence to Nvidia’s lagging stock chart, the most important sign may be in the bill of materials, not the stock price: every rack still needs a GPU, and every data center still needs reliable power.
