Central processing units (CPUs) and graphics processing units (GPUs) are the pillars of enterprise AI infrastructure in 2026, and for good reason.
CPUs are the “main brain” of a computer, handling general tasks one by one, and a GPU is a specialized chip with thousands of smaller cores built to process massive amounts of data at the exact same time (a technology called parallel processing).
[]
On Wall Street in 2026, investors continue to snap up shares of CPU and GPU stocks, mainly because the explosive growth of has created record-breaking global demand for high-performance microprocessors, and both chip types certainly qualify. On the AI front, CPUs still run operating systems, databases and enterprise applications, and coordinate AI workloads, while GPUs perform the parallel processing required to train and run AI models.
There’s a third essential slice of the chip market, too: Custom AI accelerators and networking chips developed by Broadcom Inc. (ticker: ) and others are silicon solutions for hyperscalers that optimize performance and reduce costs, both appealing prospects to already budget-strapped AI developers.
Taken together, these market-leading chip channels are driving heavy demand in AI infrastructure, data centers and advanced processors. The big question for investors is how to tell whether companies like Nvidia Corp. (), Advanced Micro Devices Inc. (), Broadcom or Arm Holdings PLC () still have room to grow versus simply benefiting from the AI enthusiasm that’s already reflected in their share prices.
Market experts believe investors must look at these companies on an individual basis when answering the query, as valuations are very different.
“Using calendar 2027 consensus EPS estimates, NVDA trades at 15.8x, AVGO trades at 18.5x, AMD trades at 33.1x and ARM trades at 82.1x,” Matt Stith, director of equity research at Cincinnati-based Bartlett Wealth Management, said recently.
Based on price-to-earnings (P/E) ratios, Stith says, the expectations are very different from company to company. He says NVDA and AVGO are not currently riding AI hype as much. “On the other hand, given where the multiples are, expectations and enthusiasm are much higher for AMD and ARM,” Stith says.
How does it all shake out for AI semiconductor stocks? These five CPU and GPU stocks should hold their own in the tech segment of an investment portfolio:
| CPU/GPU STOCKS | MARKET CAPITALIZATION* | EV/EBITDA** |
| Nvidia Corp. () | $4.7 trillion | 28 |
| Broadcom Inc. () | $1.8 trillion | 44 |
| Intel Corp. () | $465.3 billion | 30 |
| Advanced Micro Devices Inc. () | $700.4 billion | 105 |
| Arm Holdings PLC () | $249.7 billion | 214 |
*As of July 30.
**Source: Finviz. Represents enterprise value to earnings before interest, taxes, depreciation and amortization.
Nvidia Corp. ()
For much of the 2020s, Nvidia has dominated the data center and AI graphics processing unit market with its advanced architecture and CUDA software ecosystem. Trading at about $197 per share and up a tepid 2% this year, all that success has turned Nvidia into the king of cash, producing roughly $158 billion in free cash flow in the past two fiscal years.
While market watchers say NVDA remains a stellar stock for the long haul, there are factors that could bear down on unsuspecting investors, and that goes pretty much for all chip stocks. “Beyond revenue growth, I want to see gross margins holding or expanding, because that tells you whether a company has pricing power or is just riding volume,” says Jenna Lofton, founder of StockHitter.com and a Wall Street-trained investment strategist.
Lofton also watches customer concentration closely, as companies like Nvidia and Broadcom both lean heavily on a handful of hyperscaler customers. “Consequently, any pullback in Google, Meta or Microsoft’s capex plans hits these stocks harder than people expect,” Lofton says. “Data center market share matters too, but margins are the tell that separates a real moat from a temporary tailwind.”
Contrary to popular opinion, Lofton tells investors not to just buy Nvidia and call it quits. “A more diversified approach is one dominant GPU or ASIC name like Nvidia or Broadcom, paired with a foundry play like TSMC, since basically everyone in the industry needs their manufacturing capacity; and then a memory name like Micron, since AI chips are driving serious demand for high-bandwidth memory,” Lofton says. “If you want simpler exposure without picking winners, something like the VanEck Semiconductor ETF () gets you the whole sector in one ticker.”
Broadcom Inc. ()
Serving as an indirect play on AI rather than a direct manufacturer of standard GPUs, Palo Alto, California-based Broadcom is up 7% so far in 2026. Specifically, the company co-designs custom AI chips (ASICs) directly for hyperscalers like Google, Meta Platforms Inc. (META) and OpenAI.
The company made big news this week with a new $200 billion strategic partnership deal with Samsung Electronics Co. Ltd. (005930.KS) that J.P. Morgan estimates could be valued at $1 trillion in future revenues for AVGO. The deal is expected to bring the two brands together on next-generation high-bandwidth memory products for .
While Broadcom doesn’t receive as much investor focus as Nvidia, larger investors have increasingly recognized AVGO as one of AI’s biggest beneficiaries, as its main business is designing custom AI chips and mastering AI data-center connectivity. Many Wall Street analysts view Broadcom as one of the premier “second-wave AI” investments because nearly every large AI data center depends on high-speed networking in addition to GPUs.
The finances back that sentiment up. Second-quarter results exceeded expectations, and future guidance suggests $29.4 billion in Q3 revenue, up 84% year over year. Meanwhile, AVGO shares are up 25% over the past year, and consensus analyst sentiment sees an average 12?month target price around $516, indicating a 35% upward swing in stock value, according to ratings site TipRanks.com.
In particular, Morgan Stanley’s Joseph Moore taps Broadcom as a clear AI favorite, citing an expected 80% market share in Google’s TPU chips as a key factor.
[Read: ]
Intel Corp. ()
Why not Intel? Trading at $92 per share and up 122% year to date, Intel is one of the tech sector’s biggest success stories in 2026. Q2 earnings were twice as high as consensus analyst estimates, a remarkable result. For starters, the CPU company reported second-quarter adjusted earnings of 42 cents per share on revenue of $16.1 billion. That doubled FactSet consensus estimated earnings of 22 cents per share on revenue of $14.4 billion. Intel has come a long way in a short time; at this point in 2025, Intel noted a loss of 10 cents a share on revenue of $12.9 billion.
While there are valid concerns over expected 2026 capital spending, Intel’s C-suite recently reported an adjusted $20 billion heading out the door, over previous estimates of $18 billion, as Intel fights to keep up with demand in the AI chip race.
“Intel is making strong and encouraging progress on its 18A and 14A manufacturing technologies, but the turnaround remains a multiyear story,” says Hendi Susanto, portfolio manager at Gabelli Funds. “The company now needs to prove it can execute at scale with strong yields, high-volume production and high profitability.”
Intel’s outlook for the PC market was also better than expected. “Demand appears to be holding up despite inflationary component costs, and management continues to improve profitability by shifting the product mix toward higher-value offerings,” Susanto says.
Additionally, Intel Foundry has made meaningful progress in reducing operating losses. “The next major milestone is successful execution of its 14A roadmap, which will be critical to winning large external foundry customers and improving long-term profitability,” Susanto notes.
If Intel is successful in that endeavor, “investors should begin to see a clearer path toward significantly higher Foundry profitability,” Susanto adds.
Advanced Micro Devices Inc. ()
AMD is trading at $486 per share after a big pop in late July following news that it’s acquired up to 2.5 gigawatts of data center capacity through a partnership with Core Scientific Inc. (). Up 101% year to date, the Santa Clara, California-based semiconductor powerhouse has seen business boom. AMD’s latest earnings report notes a quarterly revenue rate of $10.25 billion, up from $7.44 billion the prior year. Net profits were $1.4 billion over the same period, compared to $709 million in the same quarter of 2025.
Noting that the month of July has taken more than a trillion dollars off chip market capitalization without a single datapoint showing weaker demand, Ilya Margolin, founder and capital markets analyst at Margolin Consult, cites continued growth in semiconductors and for AMD in particular. “Forward earnings multiples across Nvidia, AMD, Broadcom and Arm differ by more than five times depending on which fiscal year the estimate uses,” Margolin says. “The estimates are carrying the valuation, not the prices.”
Margolin says that AMD continues to expand with leading AI companies, which should pave the way for more growth down the road. “AMD has contracted 6 gigawatts with OpenAI and 6 (gigawatts) with Meta and issued each a (performance-based) warrant for over 160 million shares,” he says. “Premium pricing, contract pricing below a company’s existing margin structure, and volume acquired with equity all land on the revenue line at identical weight.”
Analysts largely back the stock, with a consensus TipRanks analyst review showing a $576 average price target on AMD, representing an 18% hike over AMD’s stock price as of intraday July 30.
Arm Holdings PLC ()
Up 106% year to date, U.K.-based semiconductor processing company Arm Holdings was on a roll up until June 30, and has since dialed back with a 34% share-price loss over the past month, mostly due to a dip in AI chip demand that isn’t expected to last too long. However, it clawed back 7% as of midday July 30 after reporting stellar quarterly results.
Arm’s fiscal Q1 earnings were on target, with the company reporting $270 million in profits at 25 cents per share; that’s up from $130 million at 12 cents per share in the year-ago quarter. Revenues were up 22%, to $1.3 billion, while royalties and licensing numbers were also on the rise, up 22% and 23%, respectively. CEO Rene Haas said he’s seeing robust demand for Arm’s AGI CPU, as customer demand has cleared the $2 billion mark for fiscal years 2027 and 2028, about twice as much as the company pegged only a month ago (at $1 billion).
With its big AI partners tightly in the fold (Nvidia, Amazon Web Services, Google, Microsoft and Qualcomm Inc. () are all customers) and its Neoverse data-center processor core shipments surpassing 1.5 billion, Arm continues to see nothing but green lights ahead for 2026 and the long run.
What to Watch With CPU and GPU Stocks
When evaluating CPU and GPU companies, there are multiple financial and operational metrics that matter beyond revenue growth, including data center market share, gross margins, software ecosystems, customer concentration and spending.
Investors should take note of all of them before buying.
“It starts with AI infrastructure spending; without this, the demand for CPUs and GPUs would be nothing like what we are currently seeing,” Stith says.
Beyond revenue growth, when looking at corporate profitability, Stith advises focusing on gross and operating margins. “A deterioration in these metrics, at some point, could suggest a more competitive environment or weakening demand,” he says. “As companies like GOOG, AMZN and META, for example, look to do more with internally developed silicon, it’s important to understand any changes in who the CPU and GPU companies are selling to.”
Investors tend to bite off more than they can chew by focusing only on the superstar AI stocks, and while it’s understandable, it’s not a great idea, some experts say.
“The biggest mistake an investor can make is thinking that a great company means a great buy,” says Evan Mills, associate financial advisor at Scholar Advising. “Nvidia, Broadcom, AMD and Arm may all have long AI relevance, but the question is how much of the stock price already reflects that optimism.”
As hard as it is for starry-eyed investors to admit, it gets difficult to imagine these companies posting so much gain and continuing their impressive growth trajectories. “Investors last year would have said the exact same thing, and they’ve been waiting for a big pullback that hasn’t come,” Mills says. “That might be because (company) earnings reports haven’t been as bad as people expected, but it could also just mean the optimism is still in the air, that people still believe AI is the future and these companies are the future.”
There’s a good historical example of that trend with Apple Inc. (), where many market players anticipated it beating earnings again and again. “Eventually, even when Apple just matched earnings, it came crashing down, because the optimism was already priced into the stock,” Mills says.
Mills believes it all comes down to the best ecosystem, the deepest customer relationships and the ability to keep customers on their platform rather than switching to a competitor. “More and more of these companies are going to pop up, so there’ll be more competition, and the question is how the frontrunners stay in the lead, or whether they let a newer company catch up with a better product,” Mills says.
That’s not to say these companies will fall or won’t remain at the forefront, “but it’s very rare to see an early frontrunner stay this far ahead for a long, prolonged stretch,” he adds.
Perhaps the biggest risk in the AI chip industry is a pullback in hyperscaler capex. “If Google, Amazon, Microsoft and Meta ever pump the brakes on data center spending even slightly, these stocks get hit hard and fast because so much of the growth story assumes that spending keeps accelerating forever,” Lofton says. “Export restrictions are the other wildcard; anything that changes access to Taiwan manufacturing or blocks chip sales to certain countries can reshuffle who wins overnight.”
Lofton is also eyeing custom silicon from the hyperscalers themselves: “If Google or Amazon gets good enough at designing their own chips in-house, that quietly erodes demand for the merchant chipmakers.”
More from U.S. 小萝莉影视
originally appeared on
