Nvidia’s latest product is a game-changer

Nvidia (NVDA) stock is up about 15.44% year to date, at the time of writing, Friday afternoon, May 29. Meanwhile, the SPDR S&P 500 index (SPY) is up about 11.06% in the same period.

While the stock has outpaced the S&P 500, its growth is lagging that of other semiconductor companies that are part of the AI boom.

Here are the gains other semiconductor companies achieved in the same period:Sandisk (SNDK) is up 608.5%.Micron (MU) is up 238%.Intel (INTC) is up 221.64%.Advanced Micro Devices (AMD) is up 140.23%.Marvell (MRVL) is up 138.14%.Broadcom (AVGO) is up 26.2%.

When we see the big gains achieved by some of these companies, especially Intel, things start to look strange.

Intel reported a GAAP net loss of $3.73 billion, yet it has rallied like crazy. Meanwhile, Nvidia reported very strong earnings and announced a big dividend increase, yet it dropped.

There are two reasons holding the stock back. It is already held by most institutional investors. The other reason might be the upcoming big IPOs of SpaceX, OpenAI, and Anthropic.

We need to remember that the SoftBank Group sold all its Nvidia shares in November 2025 and dumped that money into OpenAI.

Something similar may be going on now with the “hottest” IPOs this year.

Nevertheless, Nvidia is now mounting an attack that aims to turn it into an unstoppable force in the semiconductor industry.

Vera CPU opens a brand-new $200 billion total addressable market for Nvidia.

Bloomberg/Getty Images

Nvidia delivers its first Vera CPUs

Nvidia confirmed that the first Vera CPUs arrived at Anthropic, OpenAI, and SpaceXAI on May 15th. Three days later, Oracle got its units, too.

There are several reasons why this CPU launch is a game-changer for Nvidia. The company says that Vera is “the world’s first processor purpose-built for the age of agentic AI and reinforcement learning.”

What makes this CPU different is that, unlike the previous-generation Grace CPU, which was built on ARM’s Neoverse V2 cores, this one features Nvidia’s custom “Olympus” cores (also based on the ARM architecture).

The Vera CPU features 88 Olympus cores and, according to Nvidia, delivers twice the performance of the Grace CPU and is the first CPU to support FP8 precision.

Related: Nvidia CEO Jensen Huang delivers sharp message to major customer

The importance of the Vera CPU was explained by Nvidia’s EVP and CFO, Colette Kress, during the first quarter earnings call:

“Vera CPU opens a brand-new $200 billion [total addressable market (TAM)] for Nvidia, a market we have never addressed before. Every major hyperscale and system maker is partnering with us to get it deployed. We have visibility to nearly $20 billion in total CPU revenue this year, setting us up to become the [world ’s] leading CPU supplier.”

Not only is the CPU a departure from relying on ARM’s design, but it is also now aimed at capturing CPU market share from Intel and AMD through standalone sales, which makes it a game-changer for Nvidia.

The first benchmark of Vera is out, but with a big caveat

The first benchmark of Vera is already out, and it was performed by Michael Larabel of Phoronix. The specialized media outlet has been developing its Phoronix Test Suite for about 19 years.

The reason behind Nvidia’s choosing Phoronix is likely its Test Suite’s enterprise value, as it can be used to find Linux kernel regressions, and Linux is the operating system used in data centers.

The results were very impressive, but they come with a caveat. Phoronix was asked by Nvidia to only perform a subset of tests. That is to say, Nvidia must have run the test before contacting him and picked only the tests it was confident the CPU would perform well on.

To further underscore that this is highly unusual, Larabel answered in the forums that he thinks the last time such a situation happened was when he was benchmarking the original Calxeda ARM server 16 years ago. 

While it is very likely the CPU wouldn’t have performed as well if all the tests had been run, Nvidia isn’t marketing this CPU as a general-purpose CPU. The other thing we should note is that there is likely a lot of room for performance optimizations in its drivers.

Why Intel should be very worried about Vera, and AMD will probably be OK for now

Instead of going through the long list of tests, we can look at the geometric mean results.

The results showed that the Vera delivered 10% better performance than the AMD EPYC 9575F. We need to note here that the EPYC CPU has only 64 cores, vs Vera’s 88.

Additionally, while this is a high-frequency CPU, it was released in 2024. AMD is launching its Venice EPYC line, which is manufactured on TSMC’s 2nm node.

Thanks to better manufacturing node and improved architecture, it is very likely that EPYC will have the crown.

Phoronix also compared Grace CPU with Vera, and unlike Nvidia’s 2x-faster claim, it got 1.63x the performance. Regardless, it is very impressive, especially when we look at Intel; things get quite crazy. Vera crushed Intel’s Granite Rapids Xeon 6980P, which has 128 cores, achieving 1.55x the performance.

More Tech Stocks:

Bank of America resets Amazon stock forecast on key product launchBank of America resets Microsoft stock forecast after earningsBank of America resets Google stock forecast after key event

Nvidia’s new CPU might be its Apple Silicon moment in the CPU space. Apple launching the M1 chip, pressured Intel and AMD to make better CPUs, Vera will do that too. The more work Intel and AMD put into beefing up their top-tier CPUs, the better it is for consumers, as there will be a trickle-down effect.

In the case of Vera, there will be no consumer-oriented CPU, but perhaps the next generation (not the incoming one N1/N1X) of Nvidia’s laptop CPUs will feature new core designs.

What analysts think about Nvidia

In a May 25 research note shared with me, Bank of America analyst Vivek Arya and his team updated their opinion on Nvidia stock.

Analysts said they believe Nvidia is trading below its historical price-to-earnings (PE) multiples and its price-to-earnings-to-growth (PEG) ratio.

According to the team, Nvidia’s 5-year historical PE multiple is 33.6x, and its PEG ratio estimate for calendar year 2027 is 0.28x.

Stocks with a PEG below 1.0 are considered undervalued compared to their growth potential.

The team’s pro forma EPS estimates for fiscal years 2027 and 2028 are $9.09 and $13.27, respectively.

Arya reiterated a buy rating for Nvidia stock and a price target of $350, based on a 26 multiple of his estimate for price-to-earnings ratio excluding cash for calendar year 2027, which is within Nvidia’s historical forward year P/E range of 25 to 56.

Tigress Financial has Nvidia stock on its Research Focus List and in its Focus Opportunity Portfolio.

The firm’s analyst Ivan Feinsethupdated his opinion on Nvidia stock following the earnings report. 

The analyst said that record first-quarter fiscal 2027 results showed AI factory demand and rising capital returns, and that Nvidia is the main beneficiary of the AI build-out.

Feinseth reiterated a strong buy rating for Nvidia stock and raised the price target to $425 from $360.

Of 54 analyst ratings, 50 rate Nvidia a buy, and only three rate it as hold, with an average price target of $305.38, according to MarketBeat.

Downside risks for Nvidia:Slowdown in AI infrastructure spending,Share erosion due to increased competitive intensity,Supply constraints

Related: 5-star analyst resets Micron stock price target

#Nvidias #latest #product #gamechanger