Beth Kindig

Microsoft $MSFT is reportedly planning to more than triple its data center capacity from 12GW to 38GW by 2032, including owned and leased capacity but excluding rented capacity from neoclouds. $AMZN $GOOG $META $ORCL

Micron alleviated some of the typical cyclicality concern by locking in 5-year terms. Yet this creates the opposite concern — that it may not fully benefit from the boom-like nature of this cycle. Our latest analysis breaks down what is changing for $MU and memory stocks. https://t.co/lCWt8SPVt6

The I/O Fund was early to identify Bloom Energy’s $BE ability to solve the time-to-power bottleneck, with shares up 1,500% since our first entry in 2025. Today, I’m joining @cvpayne on Making Money at 2:10pm EST to celebrate $BE joining the S&P 500 and discuss what’s next for one of the I/O Fund’s biggest winners.

Apple $AAPL is also rumored to be considering using $NVDA NVLink Fusion for the servers, though plans are not finalized.

CoreWeave $CRWV announced that it has brought a multi-rack $NVDA Vera Rubin NVL72 cluster online, spanning hundreds of Rubin GPUs in a scale-out cluster.

SK Hynix $SKHY is said to be in talks with Intel $INTC over a deal to manufacture memory chips in the US for the first time, either via a JV structure or leasing part of Intel’s Ohio facility, per Reuters. $MU

Meta $META said the two chips are expected to be its ‘workhorses’ for general-purpose inference, while future generations will target improved speed and throughput. $AVGO $NVDA $AMD

The HBM market is expected to grow 25X in four years, from $4B in 2023 to $100B in 2027. Meanwhile, it’s hotly debated whether memory has topped or will continue to lead the market. This week, my free analysis tackles one of the highest-stakes questions in the AI trade. Link in bio. $MU $NVDA $AMD

Oracle: “Of all the GPUs that came up for renewal in Q1, that capacity was renewed or resold at a 20% premium to prior contracts. The majority of those GPUs are 4 years or older.” $ORCL $NVDA $AMD

Spending to secure AI systems is forecast to rise at a 64.8% CAGR through 2028, more than 5X faster than overall security spending at a ~12% CAGR. https://t.co/9rJQgo8vh4 $CRWD $ZS $PANW https://t.co/RAS9oBUdUC

Over the past three years, weekly cyberattacks per organization have surged by 70%, driven largely by an 89% increase in AI-enabled cyberattacks. $PANW $ZS $CRWD

Morgan Stanley estimates Nvidia $NVDA will consume 37.3% of total HBM demand, Google $GOOG at 36%, and AMD $AMD at 12.1%, leaving only around 15% for other players. $MU $SKHY

Within the 2030 forecast, data centers are expected to account for 83% of the total memory market or $3.05 trillion, with AI data centers accounting for more than $1.5 trillion. $MU $SKHY $NVDA $AMD

From 2025 to 2030, McKinsey forecasts that the total addressable market for sovereign AI will reach between $500 billion and $600 billion, representing 30% to 40% of overall AI demand. $NVDA $PLTR $AMD

Annual token usage is forecast to reach 4 quintillion by 2030, per Evercore, with agentic AI accounting for the majority of token generation. $MSFT $AMZN $NVDA $GOOG

In early 2025, Palo Alto saw GenAI-related data loss prevention (DLP) incidents rise from 2.5% to 14% of total DLP incidents, a more than 5X increase in just 2.5 months. $PANW $CRWD $ZS https://t.co/RDsmezL8Uf

Over the last six months, the Nasdaq CTA Cybersecurity Index has gained approximately 55%, outpacing the AI-hardware-heavy semiconductor index at roughly 42%. $CRWD $PANW $ZS

Mistral raised a record €3 billion in a new funding round with participation from Nvidia $NVDA, Samsung and others, roughly doubling the open-weight model developer’s valuation to €21 billion ($24 billion). Mistral said that its ARR recently surpassed $1 billion. $AMZN $MSFT $GOOG

OpenAI’s enterprise revenue rose approximately 32% MoM from June to July, outpacing its overall ARR growth of 20%, per CFO Sarah Friar. Enterprise now accounts for roughly half of revenue, ahead of expectations to reach that split by year-end. $AMZN $GOOG $MSFT $NVDA $AVGO

Google Cloud $GOOG CEO Thomas Kurian: “We offer 2.7x better price performance for training, 80% better price performance for inference, 30% better price performance for CPUs.” $NVDA $AMD $AVGO

OpenAI CFO Sarah Friar said that the company recently cut the price of its Luna model by 80%, which helped drive a 10X increase in usage. $NVDA $AMZN $MSFT $GOOG

CoreWeave $CRWV CEO Michael Intrator: “We are struggling to meet demand every day. Every GPU we have could be sold to multiple different clients.” $NVDA

Nvidia $NVDA is collaborating with cloud partners in Australia including IREN $IREN, Firmus, NEXTDC and others to build out 2GW of AI capacity in the country by 2027, more than doubling Australia’s current capacity of 1.6GW.

Nvidia $NVDA CEO Jensen Huang previously called Palantir $PLTR Ontology “probably the single most important enterprise stack in the world today.” Now, the relationship between the two AI leaders is expanding, and it could open the door to a much bigger opportunity for Palantir. Read our latest analysis⬇️ https://t.co/Hv56tMvgVl

Palantir $PLTR and Nebius $NBIS are also collaborating to accelerate the deployment of new compute capacity for Palantir customers, including via modular data center deployments at already-powered sites.

Revenue at the top 10 foundries rose 11.5% QoQ to $53.5 billion in Q2, with TSMC commanding 72.5% market share with revenue up 12.1% QoQ to $40.2 billion on strong AI demand. $TSM $INTC $NVDA

Google: “Our payback period on AI servers in aggregate is less than 2 years and on our own silicon is half that. So we have strong payback period.” $GOOG $AVGO $NVDA

Google $GOOG is investing at least €13 billion ($15.1 billion) to develop new AI infrastructure in Finland over the next two years, including three new data centers and a 22-year nuclear power deal. $NVDA $MSFT $AMZN

OpenAI is reportedly making progress working with Samsung on research and production of its next-generation chips, per Reuters, with this expanding beyond memory to potentially include foundry or packaging services. $NVDA $AVGO

Samsung’s first 2nm fab in Taylor is reportedly fully booked by Tesla $TSLA, Arm $ARM and Broadcom $AVGO before operations have commenced, with trial production expected this month or next and mass production in 2027. $TSM $INTC