Nvidia (NVDA) stock has hardly had a quiet 2026.
According to Seeking Alpha, shares of the AI giant are up nearly 14% year to date but have slipped around 1% over the past three months, with concerns over AI capex, China roadblocks, and memory bottlenecks keeping Nvidia stock relatively contained this year.
Also, competition is coming and putting pressure on Nvidia, but “Mad Money” host Jim Cramer dismisses the latest threat from OpenAI, arguing that investors are overlooking what makes Nvidia unusually difficult to displace.
OpenAI isn’t alone, though.
“Big Short” Michael Burry pointed to Etched as potentially “serious competition” after the AI-chip startup raised a whopping $700 million in a Jane Street-led round, valuing it at $21 billion.
Now OpenAI has added another challenge. However, in the Aug. 25 episode of “Mad Money,” Cramer’s response suggests Nvidia’s real defense might extend far beyond its chip itself.
Cramer sees little threat to Nvidia from OpenAI
OpenAI’s new Jalapeño chip is apparently giving Nvidia investors something concrete to look at.
Built with Broadcom (AVGO) specifically for AI inference, OpenAI argues that its ASIC delivered 1.5 to 1.9 times more AI work per watt compared to systems using Nvidia’s GB200 or GB300 superchips across multiple models, cutting end-to-end latency by 1.7 to 3.6 times, as reported by Tom’s Hardware.
More Jim Cramer:
- Jim Cramer has terrifying one-word message for tech stock investors
- Jim Cramer says he’s steering clear of one popular stock
- Jim Cramer reveals 4 surging chip stocks he likes best
That seems like exactly the threat Nvidia bears were waiting for. Cramer’s reaction, though, was mostly dismissive.
“Be my guest. Maybe they can,” Cramer said of OpenAI’s effort to reduce their reliance on Nvidia hardware, before questioning if that challenge may amount to “big hat, no cattle.”
His core argument centers less on whether OpenAI can design a competitive chip and more on what it would take to displace Nvidia once customers are embedded in its ecosystem. Leaving Nvidia means leaving a platform that many engineers and startups already prefer.
A custom AI chip can outperform Nvidia at specific tasks without replacing the flexibility, software, and broad ecosystem that make Nvidia the go-to AI platform.
OpenAI acknowledges that limitation.
It plans “small volumes” of Jalapeño this year, with a far bigger ramp in 2027, and explicitly states that it doesn’t expect the chip to replace its entire compute lineup. Nvidia still remains one of its “very good partners.”
Hence, replacing Nvidia across an entire AI infrastructure stack is a completely different ball game, and for now, even OpenAI is signaling coexistence rather than a clean break.
Cramer sees strategy behind Nvidia’s OpenAI bet
The irony is tough to ignore, though.
Cramer argued that Nvidia has invested a massive $30 billion in OpenAI, as reported by Reuters, even as OpenAI is building chips designed to reduce reliance on Nvidia’s hardware.
However, the veteran investor doesn’t see this as Nvidia funding its own disruption. Instead, he frames it as an ecosystem strategy, arguing that the AI behemoth can use its profits to back businesses shaping AI while remaining the go-to platform they rely on.
In other words, Nvidia becomes “the sun” in an increasingly large AI solar system. Interestingly, this comment follows Cramer’s recent hailing of Nvidia as a macro indicator that signals much more about the broader tech market’s pulse.
Moreover, Cramer dismissed concerns that Nvidia’s investments are creating “circular” demand, calling them “lazy Susan deals” while rejecting dot-com-era comparisons. Nevertheless, investors need to watch the quality of that revenue.
If Nvidia-backed companies are using Nvidia’s money to buy its chips, the cycle works while funding is strong, but it will likely weaken if that funding dries up.
Cramer actually sees bigger Nvidia risks elsewhere
OpenAI is grabbing headlines, but Cramer appeared much more concerned about risks that might constrain the size of Nvidia’s massive opportunity.
The first issue pertains to the backlash against data centers. Cramer warned that political and community opposition could slow construction, arguing that hyperscalers need a lot better engagement with local communities. He said the controversy is already weighing on Nvidia’s shares.
That threat is a lot different, as data-center restrictions, by contrast, limit total AI infrastructure spending by slowing down access to power, land, water and permits. In that scenario, the entire compute market grows a lot more slowly.
Supply constraints are another major issue.
Cramer pointed to shortages of high-bandwidth memory, which includes the DRAM Micron is looking to produce, and said those shortages are already costing Nvidia potential sales.
Though that can be perceived in a positive light (demand outpaces supply), it also shows that Nvidia is incredibly dependent on an enormous ecosystem of memory, networking, power, and data-center capacity scaling along with its GPUs.
Then there is China. Cramer noted that Nvidia has lost access to part of the massive Chinese market as Beijing continues pushing to develop its own high-end chips.
Nvidia’s pullback is resetting its valuation
For Nvidia investors, the recent choppiness is bringing the valuation a lot closer to the earnings story.
Nvidia stock recently traded near $213, 9% below its May record close of $235.47. Moreover, shares dropped as low as $197 in August before recovering, while a seven-session losing streak showed off how quickly sentiment can turn when expectations around AI spending weaken.
Nvidia stock’s valuation looks relatively less demanding now.
According to Seeking Alpha, its forward non-GAAP earnings multiple of 23.63 times is roughly 4% above the sector median of 22.66 times and around 45% below Nvidia’s five-year average of 42.85 times.
That’s a major reset for a business that’s still growing quicker than most large chip stock peers.
However, the stock isn’t cheap across every measure. Nvidia’s forward EV/sales multiple, which looks at the value of the entire business with expected sales, sits near 12.9 times, versus just 3.5 times for the sector. Also, its forward EV/EBITDA, measuring its valuation against expected operating cash-like earnings, is also roughly 31% above the sector median.
So the premium is essentially the price investors are still paying up for Nvidia’s margins, sheer scale, and AI dominance.
That said, for long-term bulls, the combo of lower multiples, powerful earnings growth, and a nearly 10% retreat from record highs creates a remarkably better entry setup if Nvidia keeps translating AI demand into earnings.