GS Johnstone GS TMT SPEC SALES ASML beats & looking to build more; Semis AI sentiment is poor; VOD appeals to Brussels against Satellites; BT

GS Johnstone - GS TMT SPEC SALES: ASML beats & looking to build more; Semis/AI sentiment is poor; VOD appeals to Brussels against Satellites; BT 14 Oct 2026 Sean Johnstone · Goldman Sachs · TMT Specialist Sales 14 Oct 2026

GS TMT SPEC SALES – Wednesday 14th October Overnight Semis lower for the 5th consecutive session (but nearly erased an intraday 3.3% drop but still de

GS TMT SPEC SALES – Wednesday 14th October

Overnight Semis lower for the 5th consecutive session (but nearly erased an intraday 3.3% drop but still declined for a fifth consecutive session, led by memory and semicap names amid renewed concerns about China’s CXMT and a broader cooling in AI bottleneck trades; MU and SNDK have now fallen in seven of the past eight sessions. In contrast, PC hardware, IT services, selected software names and the Mag 7 remained firm.

This morning will be a test just how bearish the market has become on AI and semiconductors. Sentiment has weakened amid concerns over power and funding constraints, uncertain returns, rising model competition and the risk that custom silicon reduces reliance on leading GPU suppliers. This plays against ASML Q3 where it delivered a good set of results and indicated potential upside to its longer-term unit guidance in 2028 and beyond. Early investor feedback has been mixed many wanting more colour on the guide but given ASML sell to a handful of customers, appears unrealistic. The message from ASML is we cannot meet the demand we have from customers. The other thing that will drive sentiment in the space today the Anthropic investor event in SF where there is fixation on its ARR.

ASML: Beat (look at the GM beat & this is before the recent price increases) and raise but what the market will like is it looking for further ways to raise its build capacity. Last quarter it gave guidance for FY27 low-NA EUV capacity of roughly 85 tools (with buy-side more like 90) and confirm potential capacity of more than 110 tools by 2028. Now it’s saying that “it's becoming very clear that 30% in 2027, 30% in 2028 will not allow us to meet our customer demand. So we have to do more and we are looking basically at ways, especially in 2028 and beyond, to further extend our capacity.”

Gross margin in Q3 beat (57.4%) ahead of a price increase and its now guiding Q4 higher at 57% (this was an issue yesterday at AMAT as there was some disappointment on their gross margin guide). Note of the recent price increases ASML saying it will start to become visible partly in the course of 2027. The full impact of that you will start to see in 2028 and beyond.

ASML is introduced 4Q26 sales guidance between €14.2-15.2bn , this is 3% above VA consensus at mid-point. Its Introduced 4Q26 GM guidance of 57% (street 56.7%). Taking the guide for revs, GMs and opex ASML’s 4Q26 EBIT is 7% above VA cons. Additionally, the company upgraded its 2026 revenues to €44-45bn (vs €43-45bn prior), which at the mid-point is 1% above VA cons.

The bears and there are a fair few saying there is no 2027 upgrade, seriously they announced a massive upgrade at Q2 and more EUV machines do not get built in a quarter. The other pt is just how bearish investor have become with sentiment appearing to be caught in a bearish feedback loop, with semiconductor investor discussions

increasingly focused on who can make the most negative case. So price action today going to be interesting. Note price action in AMAT was not great down 7% amid a sector-wide selloff and reflected some disappointment given the lack of a specific medium-term WFE forecast and earnings target (prior to 2030). See jim’s take and Alex ASML take

Repeated questions on AI/Semis yesterday where sentiment has turned more cautious as investors question whether the AI infrastructure build-out can meet the high expectations embedded in valuations, particularly into 2028. The debate has moved beyond the scale of long-term AI demand to the practical constraints on delivering it: access to power, the availability and cost of funding, intensifying model competition—including Meta’s newly released Muse/Wonder models—and the risk that custom silicon redistributes value across the supply chain. With positioning still…