Teclaro

𝗠𝗼𝗿𝗴𝗮𝗻 𝗦𝘁𝗮𝗻𝗹𝗲𝘆 (Stephen Byrd and a team of 20+ analysts) says the selloff in AI stocks after the US midterms is overdone, and recommends a selective "𝗕𝘂𝘆 𝘁𝗵𝗲 𝗗𝗿𝗼𝗽." 𝗧𝗵𝗲𝗶𝗿 𝗿𝗲𝗮𝗱: the elections don't kill the data center boom, they delay it. 2027 projects stay on track, but ~20% of 2028 US projects (about 5GW, mostly in Texas and Ohio) slip into 2029 as states add new requirements. That leaves US data center builds 𝗳𝗹𝗮𝘁 𝗳𝗿𝗼𝗺 𝟮𝟬𝟮𝟳 𝘁𝗼 𝟮𝟬𝟮𝟴, 𝘁𝗵𝗲𝗻 𝘂𝗽 ~𝟰𝟬% 𝗶𝗻 𝟮𝟬𝟮𝟵. Hyperscaler capex forecasts are unchanged: $𝟭.𝟲𝗧 𝗶𝗻 𝟮𝟬𝟮𝟳 𝗮𝗻𝗱 $𝟭.𝟴𝗧 𝗶𝗻 𝟮𝟬𝟮𝟴. 𝗪𝗵𝗲𝗿𝗲 𝘁𝗵𝗲𝘆 𝘀𝗲𝗲 𝗼𝗽𝗽𝗼𝗿𝘁𝘂𝗻𝗶𝘁𝘆: 1. Hyperscalers: $AMZN $META $GOOGL $MSFT, plus $NVDA 2. De-rated data center and power names: $EQIX $VST $CEG $TLN $BE 3. Large-cap semis and optical: $AVGO $TSM $LITE $COHR $ANET 4. Software and AI adopters, where they think alpha is moving next In their view, a longer cycle actually helps: less oversupply, better credit and pricier compute. Are you buying this dip?

𝗚𝗦 asked the question everyone has: how much runway is left in the US data center construction boom? Their answer: 𝗰𝗹𝗲𝗮𝗿 𝘃𝗶𝘀𝗶𝗯𝗶𝗹𝗶𝘁𝘆 𝘁𝗵𝗿𝗼𝘂𝗴𝗵 𝟮𝟬𝟮𝟴. 𝗧𝗵𝗲 𝗻𝘂𝗺𝗯𝗲𝗿𝘀: • ~𝟴𝟲 𝗚𝗪 𝗶𝘀 𝘂𝗻𝗱𝗲𝗿 𝗰𝗼𝗻𝘀𝘁𝗿𝘂𝗰𝘁𝗶𝗼𝗻 𝘁𝗼𝗱𝗮𝘆. At ~$3.5B per GW (construction only, no chips or servers), that's ~$𝟯𝟬𝟬𝗕, about 𝟲𝘅 what was spent building data centers in all of 2025. • Add the high-probability announced projects and the pipeline grows to ~$𝟵𝟬𝟬𝗕. Everything announced: ~$𝟭.𝟵𝗧. • Construction spending accelerated from +𝟱𝟱% 𝗬𝗼𝗬 𝗶𝗻 𝗝𝘂𝗻𝗲 𝘁𝗼 +𝟵𝟬% 𝗶𝗻 𝗦𝗲𝗽𝘁𝗲𝗺𝗯𝗲𝗿. • GS raised its 2026/2027 spending estimates to $80B/$100B (from $67B/$87B) and sees $𝟭𝟮𝟱𝗕 𝗶𝗻 𝟮𝟬𝟮𝟴. 𝗪𝗵𝗮𝘁 𝗮𝗯𝗼𝘂𝘁 𝘁𝗵𝗲 𝗺𝗶𝗱𝘁𝗲𝗿𝗺𝘀? GS expects moratoria in Texas, Georgia and Michigan, states with ~45% of planned capacity (Texas alone ~25%). But their read is that 𝘁𝗵𝗶𝘀 𝘀𝘁𝗿𝗲𝘁𝗰𝗵𝗲𝘀 𝘁𝗵𝗲 𝗰𝘆𝗰𝗹𝗲 𝗿𝗮𝘁𝗵𝗲𝗿 𝘁𝗵𝗮𝗻 𝘀𝗵𝗿𝗶𝗻𝗸𝗶𝗻𝗴 𝗶𝘁: projects already being built don't stop, and timelines were already too aggressive. What they can't see clearly is 𝟮𝟬𝟮𝟵-𝟮𝟬𝟯𝟬. That depends on announcements turning into actual construction. For builders like $FIX and $EME , bookings lag project starts by about 6 months, so they have a cushion if things pause. 𝗡𝗲𝘅𝘁 𝘁𝗵𝗶𝗻𝗴 𝗜'𝗺 𝘄𝗮𝘁𝗰𝗵𝗶𝗻𝗴: the Texas data center audit, due December 10.

Deutsche Bank: “After The Plunge.” El posicionamiento en acciones se desplomó a underweight en una de las peores semanas de la historia (-0.54sd, percentil 20). • Discrecionales: -1.1sd, percentil 3 (piso histórico) • Sistemáticos: bajaron a neutral (0.1sd, percentil 44) • Tech: debajo de los mínimos de Liberation Day, y ya bajo el resto del S&P El SPX ya descuenta crecimiento de earnings en cero. Tech, crecimiento negativo. Eso con Q3 de Tech al +53%. DB lee la caída más como unwind forzado de posiciones que como “blue sweep”. Y se mantienen overweight Tech: el resorte está más cargado. Paradoja: en la misma semana, los fondos de acciones recibieron US$89.2bn — el mayor inflow en 20 semanas. $SPY $QQQ $XLK