The next leg of the AI trade runs through late August and early September, when three of the largest AI infrastructure names step up to report. NVIDIA (NASDAQ:NVDA | NVDA Price Prediction) prints results on August 26, 2026, Marvell Technology (NASDAQ:MRVL) follows on August 27, 2026, and Broadcom (NASDAQ:AVGO) closes the trio on September 2, 2026. Each has already telegraphed accelerating data center demand, and each carries a specific catalyst that could reset the AI capex narrative for the back half of the year.
The setup matters. Guidance from last quarter across all three points to sequential acceleration, and prediction markets are pricing in strong follow-through. Below is the case for each name heading into the earnings report.
NVIDIA: The Anchor of the AI Capex Cycle
NVIDIA enters the quarter with unmatched operating leverage. Last quarter’s revenue hit $81.615 billion, growing 85.23% year over year, while non-GAAP EPS of $1.87 beat consensus by 5.42%. Data Center revenue alone reached $75.246 billion, up 92%, with networking accelerating 199%. Guidance for the current quarter calls for $91.0 billion in revenue at a 75.0% non-GAAP gross margin.
CEO Jensen Huang framed the moment plainly: "The buildout of AI factories, the largest infrastructure expansion in human history, is accelerating at extraordinary speed." Capital returns back that confidence. The board approved an $80.0 billion repurchase authorization and lifted the dividend from $0.01 to $0.25 per share.
Shares closed at $217.50 on August 11, 2026, up 16.76% year to date and 996.42% over five years. Polymarket gives a 95% probability that Q2 Data Center revenue tops $80 billion and an 89% probability that non-GAAP gross margin lands in the 74% to 76% band.
Risk: Guidance assumes zero Data Center compute revenue from China, and H20 shipments to China were zero last quarter. Any policy shift, positive or negative, would be the swing factor.
Broadcom: The Custom Silicon Alternative
Broadcom’s AI story has quietly become one of the most credible growth arguments in the group. Last quarter, revenue climbed to $22.19 billion, up 47.9%, with AI semiconductor revenue of $10.80 billion growing 143%. That was the eighth consecutive EPS beat.
CEO Hock Tan pointed to a bigger step function ahead: "In Q3 we expect semiconductor revenue from AI to grow over 200 percent year-over-year to $16.0 billion." Total Q3 revenue guidance sits near $29.4 billion, or roughly 84% growth, with adjusted EBITDA near 68% of revenue.
The stock closed at $416.08 on August 11, up 20.66% year to date and 37.92% over the last year. Polymarket assigns a 91% probability that Q3 AI revenue clears $15 billion and 73.5% for topping $16 billion, meaning a print at or above management’s number would still be treated as a win.
Our own podcast team framed the setup this way: "We believe that we could see a significant acceleration in the back half of 2026. And for Broadcom, that’s a little more specific to probably some chip orders from companies like Google and Meta."
Risk: Customer concentration is real. A limited number of hyperscalers drive the AI segment, and the VMware deal left the balance sheet more levered than in prior cycles.
Marvell: The Highest-Torque Name on the List
Marvell is the smallest of the three and, by year-to-date performance, the most explosive. Shares finished at $212.31 on August 11, up 150.20% year to date and 175.32% over the past twelve months. That kind of run has cooled recently, with the stock down 9.97% over the last month, which sets up an interesting risk/reward into the earnings report.
Fundamentals justify the move. Q1 revenue reached $2.418 billion, up 27.6%, with the Data Center segment at $1.833 billion and 76% of the mix. Free cash flow of $483.1 million grew 126.81%. Q2 guidance calls for revenue of $2.70 billion at the midpoint, implying roughly 35% growth, and non-GAAP EPS of $0.93.
CEO Matt Murphy was direct: "We are seeing exceptional AI-related bookings, and as a result, we are significantly raising Marvell’s revenue outlook for both fiscal 2027 and fiscal 2028." The product roadmap now stretches across 800G and 1.6T scale-out optics, 51.2T Ethernet scale-out switches, and custom XPU silicon, further extended by the closed acquisitions of Celestial AI and XConn Technologies.
Risk: The customer list is short, hyperscalers can in-source custom silicon, and a $331.8 million contingent consideration charge in the quarter is a reminder that acquisition accounting is still moving through the P&L.
What to Watch Next
The through-line across all three names is the same: AI infrastructure spending is compounding faster than the sell-side modeled six months ago. Global chip industry revenue reached $796 billion in 2025, a 26% year over year jump driven by data center and AI demand. Keep an eye on the stock reactions to guidance rather than the headline beats. If NVIDIA’s next quarter guide clears $95 billion, Broadcom reaffirms its $16 billion AI number, and Marvell raises again, the trade extends. If any of the three flinches on guidance, that will tell investors more about the durability of the cycle than any beat.
Contact [email protected] for any questions or corrections.