Alex Thompson

$SPCX is in talks to provide computing power to the Pentagon. The potential agreement could be worth “several billion dollars.”

$NBIS secured its first senior debt facility for ~$775M. Nebius will use the proceeds to “accelerate the global buildout” of its full-stack AI cloud platform.

Good Morning.

$META is hiring longtime $AMZN AWS infrastructure executive Dave Brown to help lead its AI data center buildout as it weighs a cloud push. Zuckerberg says Meta cloud is “definitely on the table” with companies seeking access to its AI models and spare compute nearly every week.

$MSFT CEO Satya Nadella criticizes Anthropic’s Fable 5 for refusing random requests saying it feels “editorially controlled” and “doesn’t make sense.” He also warns against AI concentrating in two companies with “token capital” while everyone else is forced to rent it.

Michael Burry says Hong Kong is worth “bargain hunting” after lagging during this year’s global AI rally. He sees a particularly attractive setup as “the shine comes off Korea, Japan and the semiconductors.”

$TSM is expected to nearly double its earnings over the next three years. TSMC does not need to predict whether the next major winner is $NVDA, $AMD, $AVGO, a hyperscaler’s custom accelerator or a new CPU architecture because nearly every contender depends on its advanced…

$OKLO is now in a nearly 80% drawdown over the past year 😳

THE FUTURE IS ON SALE RIGHT NOW Drawdown from highs: • $OKLO−76% • $NVTS−63% • $CRWV−60% • $ASTS−58% • $IONQ−57% • $PL−57% • $AAOI−56% • $RKLB−55% • $IREN−54% • $ONDS−53% • $TE−51% • $OUST−42% • $ARM−42% • $BE−39% • $NBIS−38%

$INTC and $GOOGL expand their strategic partnership to deploy Gemini Enterprise across Intel and accelerate its AI-driven transformation. The deal deepens Intel's use of Google Cloud as it modernizes operations with enterprise AI.

$MRVL is expected to nearly triple its revenue over the next three years, Its one of the few names spanning both sides of the AI stack with custom silicon on the compute side and switching, electro-optics and high-speed connectivity on the data movement side.

The AI economy is reorganizing the world's most important chip manufacturer in real time as HPC reaches 66% of $TSM revenue. Even more impressive this record came with 2nm contributing just 3% of wafer revenue which means TSMC’s largest node transition has barely begun.

$MU signs a long-term supply agreement with $QCOM to provide memory and storage for AI-enabled vehicle platforms. The deal supports infotainment ADAS and connectivity while giving Qualcomm greater supply and pricing visibility.

$BE is partnering with Industrial Development Funding and Oaktree on a $1.7B investment to deploy fuel cells for $NBIS AI cloud infrastructure. The project will provide dedicated behind-the-meter power for AI compute with Morgan Stanley supporting the financing.

Elon Musk says $SPCX AI satellites will use batteries to boost peak power from 150 kW to 230 kW. The added power could reduce the number of satellites needed to reach the same compute capacity while lowering launch and manufacturing requirements.

$AAPL on-device AI has reportedly received approval for iPhones in China. $BABA says Qwen will power Apple Intelligence across iOS, iPadOS, macOS and visionOS for Chinese users.

$INTC 18A yields reportedly improved to 85% (from 65% last quarter) bringing the process close to $TSM N2 and well ahead of Samsung SF2. The progress could allow Intel to insource 80%–90% of Nova Lake tiles through IFS instead of outsourcing most production to TSMC.

$ASTS partner Bell Canada has completed its first sovereign direct-to-device satellite ground station and begun testing it with the AST SpaceMobile constellation. The milestone advances the infrastructure needed to launch satellite-to-phone service across Canada.

$KTOS secured roughly $400M in Pentagon funding to accelerate hypersonic and national-security programs. $RKLB is a direct Kratos supplier so more vehicle development could translate into additional HASTE launch orders.

$AAPL is evaluating PrismML technology that can compress powerful AI models enough to run directly on an iPhone. The breakthrough could make on-device AI faster, more private and less reliant on costly cloud infrastructure.