By Evan Coffey on September 3, 2026September 3, 2026 By Evan Coffey It’s worth asking what a good quarter is actually worth right now when assessing the AI trade. Nvidia reported earnings that handily beat expectations, results strong enough to make technology the only S&P 500 sector to close higher that day, up 3.16%. By that measure, a good quarter still moves markets. But that strength hasn’t been showing up consistently across the rest of the AI trade for months, and Nvidia’s report was just most recent example of a pattern that’s been building. The full trading week ending on August 28th told a narrower story. The S&P 500 gained just 0.49% and the Nasdaq 100 added 0.43%. Under those modest headline numbers, only 3 of the 11 S&P 500 sectors finished higher. Breadth was thin even in a week when the market was in the green. Then, on Friday, Fed Chair Kevin Warsh’s hawkish tone in Jackson Hole pulled the Nasdaq down 0.50%, a reminder that the AI led strength sitting on top of the index could still be undone by a few remarks from the Fed. The gap between Nvidia and the rest of the market points to something more than a lack of breadth. It’s starting to feel as though Nvidia is breaking correlation with the rest of the AI trade. The chipmaker delivered the kind of quarterly results that in past cycles would have carried the whole trade higher with it. This time, the rest of the group barely moved, or moved in the entirely wrong direction. Nvidia proved that the demand for their chips is as alive as ever. The rest of the trade has yet to prove it can capture that demand the same way. The Results Nvidia reported fiscal second quarter results that blew past Wall Street expectations. The chipmaker posted revenue of $96.22 billion, up 106% year-over-year. Adjusted earnings per share came in at $2.22, versus estimates of $2.10. Data center revenue reached roughly $89 billion, with hyperscale revenue up 102% and AI cloud and enterprise revenue up 138% year-over-year. Guidance was just as notable. Nvidia pointed to next quarter sales of $108 billion, above the $104.2 billion analysts were projecting, even while noting that outlook assumes no data center sales from China. Demand extended beyond Nvidia’s own numbers. Amazon Web Services agreed to buy 2 million Nvidia GPUs alongside Nvidia’s new Vera CPU, and Nvidia disclosed that capital spending among the top five hyperscalers is expected to rise to $1.3 trillion next year from approximately $1 trillion in 2026. Following the report, the market’s reaction broke a pattern. Nvidia shares had fallen the day after each of its previous four earnings reports despite beating estimates each time. This time, shares rose nearly 9%, adding roughly $440 billion in market capitalization in just a single session. Peer Analysis As it has been in the past, Nvidia’s report was a continued preview of hyperscaler capex commitments, data center growth, and AI demand that should have applied just as well to the rest of the chip and cloud ecosystem tethered to the same buildout. The question following Nvidia’s earnings report was whether the market would treat their results as a rising tide or as a bar that only Nvidia itself could clear. Marvell Technology offers the clearest answer to this question. Reporting the day after Nvidia, Marvell posted record quarterly revenue of $2.74 billion, beat earnings estimates, and raised its fiscal 2027 and 2028 revenue guidance. The results echoed Nvidia’s own demand signals almost point for point. By every operational measure, the quarter was strong, yet the stock fell more than 10% the day following the report. Other chip and interconnect names like Coherent and Credo Technology slipped in sympathy with Marvell’s post earnings reaction despite carrying little of Marvell’s specific guidance risk. AMD didn’t even need an earnings report to show the same split. Shares are up over 120% for the year yet fell more than 14% in the month leading to Nvidia’s print, even as the broader AI capex narrative Nvidia would go on to confirm was building the entire time. The pattern across these names is consistent. None of them delivered numbers dissimilar to Nvidia’s message about AI demand. What changed was how much confirmation the market required before rewarding the stock. Expectations & Proximity The most straightforward explanation for this bifurcation is expectations, not true fundamentals. Marvell, AMD, Credo and the rest of the group had already priced in the assumption that everything would go right. This meant that numbers that matched Nvidia’s stellar report card still left room to disappoint on the margin, on timing, or on guidance that merely met rather than exceeded an increasingly high bar. Nvidia, by contrast, had spent four straight quarters selling off after beats. This pattern trained the market to expect a letdown even from good numbers, which paradoxically lowered the bar. There is a second explanation sitting underneath the first, and it has less to do with sentiment. Instead, it comes down to where the company sits in the AI value chain and how close it stands to the hyperscalers. Taiwan Semiconductor exemplifies this point on the positive side. As the foundry that manufactures Nvidia’s chips, TSMC’s revenue scales in near lockstep with Nvidia’s own production volumes. Shares are up roughly 38% this year and extended their rally past $420 in the days following Nvidia’s print. There is no separate customer relationship to ramp and no independent timeline to prove out. When Nvidia sells more chips, TSMC’s revenue reflects it. Marvell Technology’s situation shows the exact opposite dynamic. On August 19th, Marvell disclosed an expanded custom chip agreement with Google, worth up to $120 billion in cumulative revenue through fiscal 2033. Additionally, Marvell handed Google a warrant for up to 58.97 million Marvell shares as an incentive tied to that spending. The deal drove shares up nearly 10%, but the warrant’s vesting structure told a more patient story than the headline did. Nearly all of it vests in tranches tied to $500 million blocks of custom product revenue booked from August 1, 2026 forward, meaning the measurement period had barely begun by the time Marvell reported its quarterly results a week later. On the earnings call, CEO Matt Murphy called the deal “game changing” and said the real revenue upside was weighted toward fiscal 2029 and beyond, not the current guidance. Investors initially cheered a headline number describing the next six years, then sold the stock when the very next quarterly report didn’t reflect any of that upside. Credo sits even further from Nvidia’s position. It doesn’t design compute silicon at all, it makes the physical interconnects, active electrical cables, and optical links that move data between chips and servers inside data centers. Its revenue depends not on Nvidia’s volumes directly but on the pace at which hyperscalers build out the racks Nvidia’s chips sit in. Put together, my hypothesis is that the split is not really about Nvidia versus everyone else. It’s about proximity. TSMC sits close enough to Nvidia’s own volumes to be rewarded the same way Nvidia was. Marvell and Credo sit on the opposite side of the fence, where the same AI demand has to pass through a customer’s separate timeline before it shows up as revenue. The market has stopped giving that gap the benefit of the doubt. A Familiar Pattern This isn’t the first time one company has confirmed a theme while its ecosystem lagged behind. The clearest parallel is Cisco during the buildout of internet infrastructure in the late 1990s. Cisco sold the routers and switches that carried internet traffic, and as demand for bandwidth exploded, Cisco’s revenue captured that demand about as directly as a company could. Dozens of smaller networking and telecom equipment names rode the same theme without sitting nearly as close to the actual growth. They were supplying components, services, or capacity one or two steps removed from where the demand was. When the buildout slowed, Cisco held up longer than the periphery before eventually rolling over itself, but the names furthest from the core were the first and hardest hit. That history points to the same fork sitting in front of Nvidia’s supply chain today. A gap this wide between a leader and its ecosystem tends to resolve one of two ways. Either the surrounding companies eventually produce their own proof and the group re-rates together, or the proximity advantage compounds and the distance between the core and the periphery keeps growing. Cisco’s own eventual decline is the reminder that proximity to demand doesn’t make a company immune, it only buys more time before the same scrutiny catches up. Which path this cycle takes is still an open question, though the difference this time is that the demand signal itself, chip volumes, hyperscale capex, data center buildout, is not in dispute. What’s changed is only how much time the market is giving the other names in the ecosystem to catch up. On the Horizon Today will offer a clean read on the question this piece has been circling, with Broadcom set to report earnings. It sits closer to Nvidia on the value chain than Marvell does, with less of the customer specific timing risk that dragged Marvell’s stock down, but it’s still reporting into a market that has shown it will punish good numbers. A strong print that holds up would argue the market is starting to distinguish between names on fundamentals rather than treating the whole group as guilty by association. Another sell off on clean numbers would argue the opposite, that the bar has moved for the entire ecosystem around Nvidia. Beyond today’s report, a few things are worth keeping an eye on. Hyperscaler capital spending guidance, especially any commentary from Amazon, Microsoft, Google, and Meta on 2027 budgets will show whether the demand signal Nvidia and TSMC confirmed is actually translating into firm order further down the chain. Custom silicon programs like Marvell’s Google deal are worth revisiting each quarter specifically for revenue recognition, since that is the mechanism that drove the gap between the headline and stock reaction. The distinction that matters going forward isn’t whether AI demand is real, Nvidia and TSMC have both made that case. It’s whether the rest of the chain can start showing the market its own direct line to that demand, instead of being judged on a customer relationship or a buildout timeline it doesn’t fully control. Evan Coffey | Investment Analyst, Asset Management Evan is an Investment Analyst on GVA’s Asset Management Team. He specializes in investment research and has a passion for finance and economics. Evan uses his strong work ethic and financial literacy to help manage GVA’s multi-asset portfolios, mitigate risk, and help coordinate wholesaler relationships. In the Spring of 2024, Evan graduated as valedictorian from Ursinus College with a BA in Finance and a minor in Management Studies. Evan took part in Finance Scholars and was a CEO of the Ursinus College Investment Management Company. Evan was also inducted into Phi Beta Kappa and Omicron Delta Epsilon during his time at Ursinus. Outside of his coursework, Evan was captain of the men’s golf team and worked on a year-long honors project focusing on the stock-bond correlation and how macroeconomic variables affect the relationship between the two asset classes. Prior to joining GVA full-time, Evan interned with GVA’s Asset Management team! During his time as an intern he focused on equity research, and shadowing wholesaler and advisor meetings. Under the guidance of the GVAAM team, Evan gained quality professional experience in the finance and asset management industry. Outside of Finance and Economics, Evan enjoys golfing, watching Formula 1 racing, going to Sixers games, and traveling with close friends and family. Disclosure: I/we have no stock, option, or similar derivative position in any of the companies mentioned, and no plans to initiate any such positions within the next 72 hours. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it. I have no business relationship with any company whose stock is mentioned in this article. Investment advice offered through Great Valley Advisor Group (GVA), a Registered Investment Advisor. I am solely an investment advisor representative of Great Valley Advisor Group, and not affiliated with LPL Financial. Any opinions or views expressed by me are not those of LPL Financial. This is not intended to be used as tax or legal advice. All performance referenced is historical and is no guarantee of future results. All indices are unmanaged and may not be invested into directly. Please consult a tax or legal professional for specific information and advice. LPL Compliance Tracking #1169500 The economic forecasts set forth in this material may not develop as predicted and there can be no guarantee that strategies promoted will be successful. All investing involves risk including loss of principal. No strategy assures success or protects against loss.