
Investing Research on AI Infrastructure, Disruptive Themes, & Growth Stocks | CEO @investmattallen
My Research 👇
Joined January 2020
- Tweets23.1K
- Following143
- Followers91.7K
- Likes69.8K
Pinned Tweet
$NVDA Nvida wrote a $1 billion check to a forgotten company.
Almost nobody understands why.
Here’s what the market is missing:
investmattallen.substack.com…
🤖 Made with AI
$CRM agreed to acquire Braze for $5.66B in cash, offering $47.50 per share to strengthen its customer engagement and real-time personalization capabilities.
The deal gives Salesforce another way to monetize AI by helping businesses turn customer data into targeted campaigns. Revenue growth from cross-selling will determine whether the acquisition justifies its price.
BeanResearch retweeted
These 6 Stocks Under $30 Could Create Generational Wealth Over the Next Decade:
1. $MARA MARA Holdings
MARA started as a Bitcoin miner and is turning itself into an AI and energy infrastructure company.
It's buying a 505 MW gas power plant in Ohio for $1.5 billion, with over 1 GW of potential capacity on the site.
Most of the AI trade is fighting over chips. MARA is buying the one input nobody can print more of, power.
BeanResearch retweeted
The U.S Government Owns A Stake In 35 Companies
Here’s The Full List:
Chips & quantum
Aeluma $ALMU: $30M
Atom Computing: $100M
Diraq: $38M D-Wave
$QBTS: $100M
Extropic: $75M
GlobalFoundries $GFS: $2B
IBM $IBM: $1B
Infleqtion $INFQ: $100M
Intel $INTC: $8.9B
I-Pulse: $250M
Kepler Computing: $245M
Multibeam: $140M
OBSIDIA: $34M
PsiQuantum: $100M
Quantinuum $QNT: $100M
Rigetti $RGTI: $100M
SandboxAQ: $500M
Thintronics: $50M
Wolfspeed $WOLF: $1.5B loan
xLight: $150M
Mining & metals
Atlantic Alumina: $400M
Elmet Group: $450M
Hertha Metals: $65M
Korea Zinc: $1.4B
Lithium Americas $LAC: 5% stake + 5% of its mine
MP Materials $MP: $400M
ReElement: $80M loan
Strategic Bauxite: $85.5M
Trilogy Metals $TMQ: 10% stake
USA Rare Earth $USAR: $277M
Vulcan Elements: $50M + $620M loan
Defense, energy & oil
L3Harris $LHX: $1B
North American Blue Energy: 35% of a Venezuelan oil company
US Steel: golden share
Westinghouse: 20% of profits if it IPOs above $30B
🚨BREAKING:
Anthropic S-1 shows $80.7 Billion in net losses during H1 2026
BREAKING: Anthropic is reportedly negotiating a $45 billion computing power deal with Nvidia-backed data center developer Nscale, according to Bloomberg.
The agreement could be finalized within days and strengthen Nscale’s prospects for an IPO.
🚨BREAKING:
SPACEX $SPCX JUST REPORTED Q3 EARNINGS
- Revenue: $14.6B (+161% YoY)
- EPS: $0.35
- Net Income: $4.4B
- Adjusted EBITDA: $8.8B (+306% YoY)
- Operating Cash Flow: $13.1B
- Space Revenue: $1.8B (+6% YoY)
- Connectivity: $5.2B (+69% YoY)
- SI: $7.6B (+830% YoY)
$APLD says construction is progressing at Delta Forge 1, with buildings now going vertical at its 300 MW AI data center campus in Louisiana.
The update follows approximately $7 million in insider purchases by CEO Wes Cummins and President Jason Gechen Zhang.
BeanResearch retweeted
McKinsey projects the optical parts that move data inside AI data centers will fall 40 to 60% short of demand through 2027.
Here Are 7 Stocks That Could Benefit:
1. $AAOI Applied Optoelectronics
Applied Optoelectronics designs and manufactures high speed optical transceivers in Texas, with Amazon as a major customer.
As a US manufacturer in a market this short on supply, it's positioned to win orders from hyperscalers that want capacity close to home.
2. $COHR Coherent
Coherent makes the lasers and optical transceivers that move data between chips inside AI data centers using light instead of copper.
When supply falls this far short of demand, the companies that can actually produce transceivers get to decide who gets them and at what price.
3. $LITE Lumentum
Lumentum makes the lasers that sit inside high speed optical transceivers, and NVIDIA invested directly into the company to secure supply.
A transceiver can't be built without the laser inside it, so a transceiver shortage is really a laser shortage, and Lumentum makes the lasers.
4. $FN Fabrinet
Fabrinet is the contract manufacturer that builds optical transceivers for companies like NVIDIA at massive scale.
When everyone is short on transceivers, the company with the factory capacity to build them becomes one of the most valuable partners in the supply chain.
5. $MRVL Marvell Technology
Marvell designs the DSP chips inside optical transceivers that process and clean up the data as it moves through the light signal.
Every 800G transceiver that gets built needs a chip like Marvell's, so closing the supply gap means selling a lot more of them.
6. $CIEN Ciena
Cien builds the optical networking systems that connect data centers to each other across cities and continents.
AI workloads are now spread across multiple data centers, and the optical links between them face the same shortage as the links inside them.
7. $GLW Corning
Corning makes the fiber optic cable that carries light signals through data centers and across networks.
Every new transceiver needs fiber on both ends of it, so more optical connections means more Corning fiber in the walls.
$CBRE is acquiring UK-based Onnec Group in a deal valued at up to $1B, expanding its data center infrastructure services as AI drives demand for more computing capacity. The deal includes $610M upfront and up to $400M in performance-based payments.Â
CBRE expects infrastructure revenue to reach $7B by 2027. As data center construction accelerates, acquisitions like this give CBRE more ways to earn revenue from installation and critical infrastructure services, not just property transactions.
$XE is partnering with $AMZN and $AEP to explore a 960 MW nuclear project in Indiana using 12 small modular reactors, targeting the growing electricity needs of AI data centers. The project would repurpose a former coal plant site.
Amazon’s involvement strengthens the commercial case for advanced nuclear, but this remains a development project, not operating capacity.
McKinsey projects the optical parts that move data inside AI data centers will fall 40 to 60% short of demand through 2027.
Here Are 7 Stocks That Could Benefit:
1. $AAOI Applied Optoelectronics
Applied Optoelectronics designs and manufactures high speed optical transceivers in Texas, with Amazon as a major customer.
As a US manufacturer in a market this short on supply, it's positioned to win orders from hyperscalers that want capacity close to home.
2. $COHR Coherent
Coherent makes the lasers and optical transceivers that move data between chips inside AI data centers using light instead of copper.
When supply falls this far short of demand, the companies that can actually produce transceivers get to decide who gets them and at what price.
3. $LITE Lumentum
Lumentum makes the lasers that sit inside high speed optical transceivers, and NVIDIA invested directly into the company to secure supply.
A transceiver can't be built without the laser inside it, so a transceiver shortage is really a laser shortage, and Lumentum makes the lasers.
4. $FN Fabrinet
Fabrinet is the contract manufacturer that builds optical transceivers for companies like NVIDIA at massive scale.
When everyone is short on transceivers, the company with the factory capacity to build them becomes one of the most valuable partners in the supply chain.
5. $MRVL Marvell Technology
Marvell designs the DSP chips inside optical transceivers that process and clean up the data as it moves through the light signal.
Every 800G transceiver that gets built needs a chip like Marvell's, so closing the supply gap means selling a lot more of them.
6. $CIEN Ciena
Cien builds the optical networking systems that connect data centers to each other across cities and continents.
AI workloads are now spread across multiple data centers, and the optical links between them face the same shortage as the links inside them.
7. $GLW Corning
Corning makes the fiber optic cable that carries light signals through data centers and across networks.
Every new transceiver needs fiber on both ends of it, so more optical connections means more Corning fiber in the walls.
Want to become a better investor?
Join my FREE newsletter that talks about growth stocks, ai infrastructure, and disruptive themes
beanresearch.substack.com/su…
BeanResearch retweeted
Why I’m Still Bullish on AI Infrastructure Despite the Selloff:
beanresearch.substack.com/p/…
$MARA $IREN $CRWV
BeanResearch retweeted
ARK Invest’s bull case for $TSLA Tesla is absolutely massive:
-ARK’s bull case puts Tesla at roughly $5,000 per share by 2031, implying a market cap of about $20.4 TRILLION
-The bull case assumes 10 million Robotaxis deployed by 2031
-Robotaxi is the core of the model, with ARK expecting it to account for the vast majority of Tesla’s enterprise value
-ARK also assumes 1.2 million Optimus robots sold externally by 2031, making humanoid robotics another major contributor to the valuation
The most important part of ARK’s model is that the upside is no longer being driven primarily by EV sales.
It’s being driven by autonomous transportation and humanoid robotics becoming massive standalone businesses.
BREAKING: Anthropic is reportedly negotiating a $45 billion computing power deal with Nvidia-backed data center developer Nscale, according to Bloomberg.
The agreement could be finalized within days and strengthen Nscale’s prospects for an IPO.
BeanResearch retweeted
Bank of America forecasts annual humanoid robot shipments will reach 10 million units by 2035.
Here Are 7 Stocks That Could Benefit:
1. $RRX Regal Rexnord
Regal Rexnord makes the motors, actuators, and micro gears that move a robot's joints, and it's one of the few US suppliers building all three into one unit.
A single humanoid needs dozens of joints, so 10 million robots a year means hundreds of millions of actuators.
$EOSE expects Q3 revenue of $82M to $83M, up roughly 170% year over year, while gross margins improved 16 to 18 percentage points sequentially. Its backlog reached $765M as production ramped at its second manufacturing line.
The bigger opportunity is turning that backlog into profitable sales. If higher factory output lowers manufacturing costs, Eos can narrow its heavy losses and improve cash generation. If Eos can improve manufacturing efficiency as production scales, its growing backlog could support sustained revenue growth while creating a clearer path toward profitability.
$BE shares came under pressure during last week’s AI selloff, despite a major deal announced October 14: $AEP plans to invest $4B in a 702 MW fuel-cell project using Bloom’s technology to supply $GOOGL under a 20-year power agreement.Â
The selloff creates near-term valuation risk, but the Google-backed project reinforces real demand for Bloom’s power systems. With construction expected to begin in 2026 and operations targeted for late 2028, the key catalyst is converting contracted demand into revenue while broader AI sentiment recovers.
$MS estimates U.S. data centers could face a 32 GW power shortfall through 2028, leaving 34% of projected electricity needs unmet even after alternative power solutions.
That could hit memory and optical chip suppliers first if customers delay deployments or push back orders. Meanwhile, $CEG and $BE could benefit as reliable nuclear power and on-site generation become more valuable to AI operators racing to bring capacity online.
BeanResearch retweeted
The World Nuclear Association projects global uranium demand will more than TRIPLE by 2040
Here Are 7 Stocks That Could Benefit:
1. $CCJ Cameco
Cameco is the largest uranium producer in the Western world, running some of the highest grade mines on Earth in Canada.
When utilities need to lock in fuel for the next few decades, Cameco is the first call they make, and demand doubling means a lot more of those calls.


















