6.7 billion. NVIDIA also delivered a staggering second-quarter fiscal year 2027 revenue of $96.2 billion on August 26, 2026, a 106% jump year-over-year, driven by its data center segment's 117% year-over-year growth to $89.0 billion. AMD similarly reported a 50% year-over-year revenue increase to
1.5 billion for its second quarter 2026 on August 4, 2026, with its Data Center segment revenue more than doubling. Despite these robust figures, institutional block trade activity has shown nuances. For the week of September 21-25, 2026, a weekly dark pool report indicated that NVDA held within its expected trading bands, while the broader market ETFs saw significant dark pool notional flows. Furthermore, a pre-market analysis on September 24, 2026, highlighted institutional entries with a '100% putside skew' for new contracts in both NVIDIA and AMD, suggesting some smart money is deploying hedging strategies or expressing short-term bearish views. This indicates a complex institutional sentiment where strong fundamentals are balanced against potential near-term risks or a re-pricing of elevated valuations, a sentiment that contributed to the Philadelphia Semiconductor Index (SOX) falling as much as 29% from its June peak earlier in the year.
The Mechanism
The current dynamics in semiconductor equities are profoundly shaped by the structural transformation driven by artificial intelligence. Institutions are navigating this landscape using sophisticated tools like dark pools and Cumulative Volume Delta (CVD) analysis to manage exposure and execute large block trades with minimal market impact. Dark pools facilitate large, off-exchange transactions, allowing 'smart money' to accumulate or distribute positions without telegraphing their intentions to the broader market. When institutional players engage in significant block trades in these venues, it can foreshadow future price movements once that hidden liquidity eventually interacts with public exchanges. Cumulative Volume Delta, a powerful metric, tracks the net difference between aggressive buying and selling volumes over time. A divergence between CVD and price action often signals a weakening trend or a potential reversal, indicating whether buyers or sellers are truly dominating the market beyond superficial price movements. The underlying mechanism for this concentrated institutional interest is the relentless demand for AI chips. Deloitte projects that generative AI chips will comprise nearly half of global chip sales in 2026, approaching $500 billion in revenue. This surge is further evidenced by hyperscale cloud providers, enterprises, and governments collectively expected to deploy $700 billion to $745 billion in AI infrastructure capital expenditures in 2026 alone. However, this demand is not without its challenges. NVIDIA's CFO Colette Kress, in late August 2026, warned of gross margin compression due to 'severe memory pricing constraints' for Q3 and Q4, with margins expected to bottom out in the 71% to 72% range. This highlights the ongoing supply chain bottlenecks in high-bandwidth memory (HBM) and other critical components, which can affect the profitability and delivery capabilities of AI chip manufacturers.
Who is Exposed
Companies at the forefront of AI chip innovation are most exposed to both the tailwinds of demand and the intricacies of institutional positioning. NVIDIA (NVDA), as a dominant force in AI GPUs, continues to capture significant attention. Its Q2 FY2027 Data Center revenue alone hit $89.0 billion, a 117% increase year-over-year, with CEO Jensen Huang noting that 'AI infrastructure buildout is at full steam'. However, the institutional put activity observed on September 24, 2026, for NVIDIA, alongside AMD, suggests that some smart money players are actively managing risk, possibly through hedging or expressing short-term downside expectations despite the long-term bullish narrative. Advanced Micro Devices (AMD) is also a key player, particularly with its strong Data Center segment, which more than doubled its revenue in Q2 2026. The company's upcoming Helios platform is highly anticipated, with commercial shipments expected by the end of Q3 2026. Earlier in the year, AI analysis identified institutional rotation into AMD during a tariff selloff, suggesting accumulation by smart money at a discount. Broadcom (AVGO), with its focus on custom AI accelerators and networking solutions, reported Q3 AI semiconductor revenue of
6.7 billion, up 221% year-over-year, and expects Q4 AI semiconductor revenue to accelerate to 1.7 billion. While a prior market correction in June 2026 was partially attributed to 'weaker than expected AI chip guidance from Broadcom' at that time, the company's actual Q3 results and subsequent Q4 guidance released in September 2026 demonstrated robust performance in its AI segment. This highlights how market perception can shift quickly, and the importance of updated fundamental data. The entire semiconductor supply chain, from foundries to equipment manufacturers and memory providers, remains highly exposed to these dynamics, with sustained demand for advanced logic, DRAM, NAND, and HBM driving investment.
Quantitative Outlook
The market data reveals a sector in flux, where strong long-term narratives meet short-term trading realities. AMD's impressive 281.02% gain over the past year is a significant outlier, indicating strong conviction and potentially sustained institutional accumulation over that period, as hinted by earlier reports of smart money rotation. NVIDIA, despite its substantial 27.82% annual rise, has faced some daily pressure, closing at 227.21, a -0.72% change today. Broadcom, closing at 355.10 with a 1.58% gain today, shows a mixed picture over the shorter term, with its 30-day and 90-day performance down -3.54% and -5.82% respectively, even as its AI segment demonstrates phenomenal growth. The high daily volumes for NVDA (98,224,511), AMD (16,325,476), and AVGO (18,872,234) underscore the intense activity in these names. These figures, when viewed alongside reports of off-exchange block trades and institutional put activity in NVDA and AMD on September 24, 2026, indicate that while the long-term AI narrative remains bullish, certain segments of smart money are actively de-risking or hedging their positions. The broader semiconductor market is projected to reach sales between $975 billion and
.29 trillion in 2026, reinforcing the secular growth trend. Critical to the quantitative outlook will be monitoring the consistency of earnings beats and upward guidance revisions from the major players, especially given the supply chain constraints impacting memory pricing. Any signs of a material easing of these constraints or, conversely, a deepening of them, would significantly alter the picture for gross margins and overall profitability. Furthermore, close attention to the Cumulative Volume Delta in these key stocks can provide real-time insights into whether the observed price movements are genuinely supported by aggressive buying or selling, or if they represent temporary fluctuations against a stronger underlying institutional flow.
Tags: Semiconductors, AI Chips, Dark Pools, NVDA, AMD, AVGO
