Beth Kindig

Nvidia $NVDA said that with software optimizations alone, it was able to deliver a 5X increase in tokens generated in the same power envelope in just one month.

AWS is expected to ship 2.4 million of its Trainium3 accelerators this year, ramping to 3.2 million in 2027. $AMZN $NVDA $GOOG

SK Hynix CEO Kwak Noh-jung: “We forecast that next year will be the worst year in the industry's history from the supply perspective. We still forecast that customer demand will remain higher than our supply capacity even beyond 2030.” $SKHY $MU $AMD $NVDA

Just hit inboxes! My Q3 Top 20 AI Stocks report, a 90+ page dive into the biggest trends ahead for AI investors and the many lesser-known stocks leading the charge. Link in bio👇 Prior Top 15 reports identified 5 positions up 100%+ YTD and ten up 50%+ despite recent volatility.

Intel $INTC is planning to spend $5.7 billion to expand manufacturing in Ireland, with the expansion primarily geared towards its Xeon 6 server processors and future generations for AI workloads. $NVDA $AMD

Meta is reportedly in talks with Anthropic to lease $10 billion of computing power, with the deal expected to see Anthropic pay Meta in monthly increments over a two-year period. $META $MSFT $AMZN $GOOG $NVDA

Big Tech’s AI monetization equation is a balancing act between scaling AI revenue and surging capex, with one Big Tech name already seeing FCF go negative. Below, we break down the debt and equity fueling the AI buildout, and which other Big Tech companies could see FCF slip into the red by the end of the year. https://t.co/r1wmjNoTzR $GOOG $META $MSFT $AMZN

AI-optimized servers consumed around 95 TWh of electricity globally last year, with this projected to rise 84% to 175 TWh in 2026. $NVDA $AMD $AVGO

Q2 earnings is on the horizon, putting Big Tech under pressure to show the Street that record capex is translating into AI-driven revenue and profits. Below, we break down the key AI monetization signals at $GOOG, $MSFT, $META and $AMZN and what metrics investors must watch next. https://t.co/KjT6Qv6ZiB

Meta is building its first data Canadian data center in Alberta, with the 1GW facility expected to see an investment of around $9.2 billion. $META $NVDA $AMD $AVGO

Micron added $700B in market cap so far in 2026, while Big Tech added only $426B combined. The market is rewarding AI suppliers far more than the companies deploying capital. Our Big Tech earnings preview breaks down why – hitting inboxes tomorrow. $GOOG $MSFT $AMZN $META

AI chip startup Positron is reportedly looking to raise $750 million across two phases valuing the company at $5 billion, more than tripling its valuation in five months. $NVDA $AMD $AVGO

Japanese chipmaker Rapidus is reportedly looking to offer its 2nm capacity at around $18,550 - $21,635 per wafer, a significant discount to the $30,000 that TSMC is said to charge and comparable to Samsung, with production ramping in 2H 2027. $TSM $NVDA $AMD $INTC

Hyperscalers have doubled their debt loads over the last five years to fund accelerated AI investments, with Alphabet, Amazon, Meta, Microsoft and Oracle collectively adding $350 billion in debt since 2021, per Bloomberg. $GOOG $AMZN $META $MSFT $ORCL

IBM: “In the last few weeks of June, we saw clients shift their quarterly capex spend toward servers, storage, and memory purchases to secure supply-constrained infrastructure ahead of expected price increases.” $IBM $NVDA $AMD $MU

Micron $MU has raised its US investment plan to $250 billion through 2035, up from $200 billion outlined last year, with construction on its New York campus said to be 25% ahead of schedule. $NVDA $AMD

Meta $META is also aiming to deploy 7GW of capacity this year, with 1GW added in the first half, before doubling this to 14GW by the end of next year. $NVDA $AVGO $AMD

AMD $AMD announced a collaboration with AI digital infrastructure provider 5C on the development of next-gen ‘gigascale’ AI data center campuses leveraging AMD’s Helios platform. $NVDA

Meta $META is expanding its Hyperion data center in Louisiana to 5GW, taking its planned investment in the facility to over $50 billion. $NVDA $AVGO $AMD

Nvidia $NVDA says a Vera Rubin NVL72 rack paired with a Groq 3 LPX rack can deliver up to 35X higher token throughput per MW, creating up to a 10X revenue opportunity for AI factories versus Blackwell. Meanwhile, acquiring 1 TB of HBM through Blackwell GPUs costs more than 5X as much as adding 1 TB of DDR5. The I/O Fund’s latest two-part analysis breaks down why offload engines may become a key solution for scaling memory independently of GPUs. https://t.co/uPYR85mgXx

AI startup Perplexity is planning to use Nvidia’s new standalone Vera CPUs, with the company saying it found that Vera carried out agenting coding tasks 1.5X faster than traditional CPUs. $NVDA $AMD

Micron $MU broke ground on its $9.3 billion fab expansion in Hiroshima, Japan, with the first equipment for the expansion expected to be delivered in 2H 2028 to help support HBM and other DRAM memory chips. $NVDA $AMD $AVGO

Inference processing power in $NVDA Rubin is 5X higher than Blackwell, but HBM bandwidth rises only 2.8X, leaving the memory wall firmly intact. As hyperscalers move from AI buildout to AI monetization, they have to use memory more efficiently. Our latest piece breaks down why offload engines are emerging as a key solution for improving tokens per watt. https://t.co/tb9aTVB6Q2 $GOOG

SK Hynix has reportedly removed price caps in its long-term memory supply agreements, unlike Micron, which will allow spot price increases to be fully reflected in its contract pricing. $MU $NVDA $AMD

Samsung is said to be “aggressively” negotiating DRAM price hikes of up to 20% QoQ for Q3, with LPDDR hikes suggested to exceed 20% due to supply tightness. $MU $NVDA $AMD

SK Hynix raised $26.5 billion in a record-breaking ADR offering, with shares set to kick off trading on the Nasdaq. $MU $NVDA $AMD

AI data center startup Nscale has received a $900 million line of credit to help expand its data center operations in the US, Europe and Asia, following a $2 billion funding round earlier this year. $NVDA $MSFT $GOOG

Anthropic is reportedly in talks with Samsung over the manufacturing of a custom AI chip using its 2nm node and advanced packaging tech, though it has yet to move into design, testing, or manufacturing stages. $TSM $NVDA $AMD $AVGO $GOOG

Foxconn recently reported 40% growth in Q2 revenue to ~$79 billion, accelerating from 30% in Q1, with Cloud and Networking accounting for more than half of revenue driven by strong AI server momentum. $NVDA $AMD

IBM unveiled its new 0.7-nanometer chip manufacturing technology, which it says packs nearly 100 billion transistors onto a fingernail-sized surface, offering 2X the density of its 2nm chip with up to 70% greater energy efficiency and 50% higher performance. $IBM $TSM $INTC