Key Highlights

  • The chip benchmark recovered 0.44% to 10,779.46 on October 23, 2026, one day after all 30 members declined, yet its weekly result remained about 3.2% lower.
  • Roughly USD 1.3 billion entered a semiconductor fund before Michael Burry publicly agreed on October 26 with Ken Griffin’s warning that excess GPU capacity could drive rental prices sharply lower.
  • Credo lost 9.5% to a six-month low after MaxLinear targeted 20% to 30% of the market for 1.6T active electrical cables.
  • Western Digital revenue increased 55% to USD 4.38 billion, but the shares still retreated about 3% as the market questioned the lifespan of the AI investment cycle.

Did a 0.44% Rebound Settle the Debate Around SMH?

The SMH fund (NASDAQ:SMH) concentrates its portfolio in major chip designers, manufacturers and equipment suppliers. Its performance consequently tends to resemble the Philadelphia Semiconductor Index, or SOX. That benchmark ended October 23, 2026 at 10,779.46, up 0.44%. The gain followed a Thursday rout in which each of the index’s 30 constituents declined. It also left the SOX down about 3.2% for the week, so the Friday move repaired only a small part of the damage.

Individual recoveries showed buyers returning after sharp falls. Micron Technology (NASDAQ:MU) advanced 2.46% to approximately USD 877.55 after losing more than 6% toward USD 850 in the preceding session. Advanced Micro Devices (NASDAQ:AMD) added 1.74%, while Intel (NASDAQ:INTC) rose 1.72%. Micron had also entered Friday with four consecutive weekly losses. Those moves establish that investors bought weakness; they do not establish that expectations for final demand improved.

The unresolved issue is spending by hyperscalers, the large operators of cloud platforms. Their purchases influence sales across processors, memory and production equipment. Meta Platforms had just disclosed negative free cash flow, intensifying questions about how long customers can fund orders at the current rate. A pause or reduction in spending guidance would lower revenue expectations throughout the supply chain. Rate pressure sits beside that operating risk because futures placed the probability of a December Fed increase between roughly 71% and 80%.

Can SOXX Inflows Withstand the GPU-Capacity Argument?

SOXX (NASDAQ:SOXX) finished Friday 0.53% higher, although it lost more than 3% for the week after dropping almost 7% on Thursday. Daily flow figures dated October 21 and released October 23 recorded approximately USD 1.3 billion of inflows. The timing matters: this was capital committed before the most severe part of the selloff, not proof that investors bought after seeing the full decline.

The opposing case came from two prominent investors. Citadel founder and chief executive Ken Griffin said on October 21 that an AI bubble was almost certain. Michael Burry publicly endorsed the warning on October 26 in characteristically blunt terms. Griffin compared GPUs with airlines because both can be operated at very low marginal cost after the expensive asset has been installed. Under that framework, too much capacity sends rental rates down, weakens the return produced by each processor and reduces the case for the next hardware order.

Semiconductor vendors receive payment when customers order chips, rather than when installed servers are used. A capacity glut would therefore appear first as a slowing order pace. Financing may reinforce the effect: Elon Musk said debt markets had almost completely closed to AI businesses lacking actual revenue. Amazon supplied the counterargument, saying 2026 capacity was sold out and 2027 was largely sold out. The fund flow and those capacity comments show why the demand debate remains open despite the warning.

Why Did a Rival’s 1.6T Plan Hit Credo So Hard?

Credo Technology (NASDAQ:CRDO) designs connectivity semiconductors and the active electrical cables that use them. The shares closed October 23 down 9.51% at a six-month low even though the wider chip index gained. MaxLinear (NASDAQ:MXL) had said on its fiscal third-quarter 2026 call that it intended to capture between 20% and 30% of the next-generation 1.6T cable segment. The designation refers to links capable of moving 1.6 terabits each second.

Known as active electrical cables, these links, usually shortened to AECs, combine copper wiring with chips that strengthen the signal between servers and switches over relatively short distances inside AI clusters. Credo has led that niche. Such links consume less energy than comparable optical links over short runs, so the concern was not an absence of demand. It was the possibility that the next generation of purchases would be shared with another supplier.

Leadership in a quickly expanding market can command a premium valuation when investors expect dominant share and durable pricing. A credible challenger seeking as much as 30% creates both a volume threat and a pricing threat, since a hyperscaler can use a second qualified source in negotiations. Long qualification cycles and Credo’s installed position work in its favor because large customers value reliability in dense computing systems. The downside case begins if MaxLinear qualifies with an important buyer.

Why Was Western Digital Sold After a Beat and Higher Guidance?

Western Digital (NASDAQ:WDC) reported fiscal first-quarter 2027 figures on October 22, 2026. The data-storage producer generated USD 4.38 billion of revenue, 55% more than a year earlier, and earnings per share increased 161% to USD 4.64. It lifted the quarterly dividend 17%. Management’s December-quarter outlook implies about 69% revenue growth, exceeding consensus forecasts by more than 10%.

The stock nevertheless fell around 3% on October 23 and remained approximately 53% below its June level. Hard disk drives store much of the information held in cloud data centers, tying demand to the investment budgets of the largest cloud providers. Reported estimates put hyperscaler capital expenditure on course for roughly USD 800 billion this year. When investors believe a spending cycle may be near its peak, they assign a lower multiple to strong current earnings because the result may not be repeatable.

Industry supply is another variable. Additional drive capacity can weaken prices while unit demand is still rising. October 27 brings results from Seagate Technology (NASDAQ:STX) and offers the next check on enterprise-storage orders. Western Digital’s reaction therefore says less about the quarter already reported than about expectations for the spending and pricing environment that follows it.

What to Watch Next: Which Results Could Decide the Chip Argument?

The combined evidence does not produce a single answer for semiconductors. SMH’s reference index rebounded, but a roughly 3.2% weekly loss remained. The SOXX inflow showed willingness to own the sector before the rout, whereas the Griffin-Burry argument questioned the economics of a heavily built GPU base. Credo illustrated the immediate valuation cost of a potential competitor. Western Digital demonstrated that 55% growth can be discounted when markets focus on a possible capital-expenditure peak.

Published estimates place the relevant spending between approximately USD 700 billion and USD 800 billion, depending on the definition. The next sequence begins with Seagate on October 27. Alphabet (NASDAQ:GOOGL), the Federal Reserve, SK hynix and Microsoft (NASDAQ:MSFT) follow on October 28. Amazon.com (NASDAQ:AMZN) and Intel report October 29, 2026. Their guidance can support the sold-out-capacity case or strengthen the glut thesis. Spending plans are forecasts rather than promises. This discussion is informational and does not provide investment advice. The outlook remains conditional.