Beth Kindig

Micron is reportedly accelerating its HBM expansion, with capacity expected to rise as much as 60K wafers per month to focus on 12-hi HBM4. This would take Micron’s total capacity to approximately 100K wafers per month, per TrendForce. $MU $NVDA $AMD $AVGO

Large-scale data centers are projected to potentially triple by 2030, rising from 90 100MW+ data centers today to 280 by 2030 if all planned projects are completed. $MSFT $META $GOOG $AMZN $NVDA

Qualcomm $QCOM recently announced a multi-generation custom silicon collaboration with $AMZN AWS for its QTI server chips, along with optical connectivity solutions up to 1.6T. Qualcomm also offered Amazon warrants to purchase up to 25M shares with vesting linked to purchase orders up to $60 billion.

Palantir has called the AI ecosystem a self-serving “token industrial complex,” taking a direct shot at the R&D labs. The analysis below looks closer at the differences between AI demand and value creation, plus one question that sits at the center of the escalating debate between $PLTR, OpenAI, and Anthropic. https://t.co/kHAwFqm0ht

Intel $INTC recently announced that it has processed more than one million 300-mm wafers using its High-NA EUV machines from ASML $ASML, a first in the industry. For comparison, TSMC $TSM is expecting to first adopt High-NA EUV machines by 2030.

Oracle: "We delivered 850 megawatts of AI capacity containing more than 300,000 GPUs to customers since the end of Q4. Delivery in Q1 is almost 3x what we delivered in all of Q4 and 73% of the total capacity we delivered last fiscal year." $ORCL $NVDA $MSFT $GOOG $AMZN

Tokens processed per day are now tracking above 400 trillion, up more than 185% since October 2025 and over 4,250% versus October 2024. But demand isn't the same as profit. One company is turning that demand into real margins while the frontier labs burn cash. This week’s analysis hits inboxes on Friday. Link in bio.

At $6 million, the average cost of an AI-enabled breach is roughly 20% higher than the global average breach cost of $4.99 million. $CRWD $PANW $ZS

Nvidia $NVDA has stated Vera Rubin can deliver 50X more throughput per MW and 35X lower token cost relative to Blackwell Ultra, and in the event that higher throughput can rise faster than TCO, customers will be more willing to pay higher prices. https://t.co/OoX2AwInm0

TSMC $TSM reported record August revenue of NT$514.8B ($16.35B), up 53.3% YoY and 10.1% MoM. YTD revenue totaled NT$3.39 trillion, up 39.3% YoY on strong AI demand. $NVDA $AVGO $AMD

As a result, the combined output of Nvidia’s $NVDA GB300, VR200 and VR300 NVL72 racks is forecast to exceed $710 billion in 2027, up 214% YoY, aided by the ASP uplift with Rubin alongside the growth in rack shipments. $AMD $AVGO

Dell $DELL on supply constraints: "The constraints remain the same. DRAM, DRAM, DRAM, followed by NAND, NAND, NAND. We have spotty CPU shortages. There are shortages with disk drives. If you go further down in the supply chain, just about every product going through a leading node is constrained." $NVDA $AMD $AVGO $MU

Memory capex is forecast to be $97 billion in 2026, rising to $146 billion in 2027 and accounting for 56% of total semiconductor capex next year. $MU $SKHY $NVDA $AMD

Estimates for TPU unit shipments range quite widely for 2027, with Morgan Stanley projecting 7.4 million TPUs shipped and UBS forecasting a more bullish 10.4 million units. $GOOG $AVGO $NVDA $AMD

Anthropic recently signed a $35 billion cloud deal with Nvidia-backed AI startup Lambda. Nvidia $NVDA will supply chips and hold the lease to the data center that Hut 8 $HUT is developing for Lambda. $AMZN $GOOG $AVGO

Considering its revenue ramp into FY28 and strong gross margin profile, a reasonable assumption to maintain a mid-40% FCF margin next year could see Nvidia $NVDA generate north of $310 billion in FCF – or enough to cover the current circular financing obligations by nearly 2X. https://t.co/38kSkpiUx0

According to Google, high-performance memory now accounts for over 75% of an AI server’s bill of materials. $GOOG $AMD $NVDA $MU

OpenAI said that its advertising business ChatGPT Ads reached a $1 billion annualized run rate, while also expanding to India, Europe, the Middle East and North Africa. $META $GOOG $MSFT

Agentic AI is creating a surprising growth outlet outside of chips and networking, with cybersecurity vendors seeing triple-digit AI-driven QoQ growth. Below, we map out why cybersecurity is a must with agentic AI, and the larger tailwinds at play. https://t.co/qWL89vz5wN $CRWD $PANW $ZS

Geothermal developer Fervo Energy signed a 396MW deal with Google $GOOG at its Cape Station project in Utah set to come online in 2028, with an option to purchase an additional 600MW, intended to support potential data center development. $MSFT $AMZN

Dell $DELL COO Jeff Clarke: "We're expecting AI to be 75% of all data center demand by 2030, adding 200 gigawatts of power over that same time frame." $NVDA $AVGO $AMD

Google $GOOG is said to be taking aim at Microsoft $MSFT and Anthropic on AI costs, introducing pay-as-you-go pricing, token discounts up to 20%, and zero-dollar base options for some customers, per CNBC. $AMZN $NVDA

Nvidia $NVDA is expanding its partnership and investing $3.5 billion into MediaTek, with the Taiwanese chipmaker joining the NVLink Fusion ecosystem, helping its custom silicon customers stand up NVLink-connected data centers. $AVGO $AMD $MRVL

AI cloud startup Lambda raised almost $1 billion in debt to purchase Nvidia $NVDA GPUs that Microsoft $MSFT will lease, per Bloomberg, with Lambda said to be expecting to quickly deploy the chips and start generating revenue to help pay back the debt.

SK Hynix $SKHY is reportedly considering Intel Foundry $INTC for a portion of its HBM4e base die production, which could help alleviate cost pressures as reports suggest TSMC’s $TSM HBM4 base dies could already cost 3–4X SK Hynix’s in-house dies. $MU $NVDA

AWS is also deploying its first AI Zone in Saudi Arabia with up to 50MW of capacity with HUMAIN in 2028, featuring both AWS' Trainium chips and Nvidia GPUs. $AMZN $NVDA $AMD

Spending allocated to AI-related cybersecurity is forecast to rise more than 5X faster than overall security spending. Below, we map out how autonomous AI agents are reshaping the cybersecurity industry – and where the fastest growth is set to hit first. $CRWD $ZS $PANW https://t.co/kiPRR1qqpn

Nvidia $NVDA is acquiring Hugging Face for $12.9 billion, though Hugging Face will remain an open platform and will not be required to use Nvidia compute.

Depending on how you model the two-quarter lumpiness, Nvidia's $NVDA ACIE can conservatively overtake hyperscalers in 5-6 quarters at a 14% QoQ growth rate versus 10% QoQ growth rate for hyperscalers, or as soon as 2 quarters if the Q2 pace continues. https://t.co/OoX2AwInm0 https://t.co/07o2BjGmP3

From 2026 to 2035, agentic AI is forecast to drive 9X enterprise traffic growth compared to just 2.5X growth without agentic AI. This week, we break down what sector is most exposed to this growth and the long-lasting tailwind this traffic growth presents.