itsbigmikey

“ "An estimated $156 billion of data center projects were cancelled or delayed in 2025, and $130 billion in 1Q26. Community pushback against data center construction has accelerated in 2026 as new moratoriums have been introduced, with the majority of the change happening at the local level. This puts pressure on costs and timelines and could alter the geographic distribution of data centers. Sustained data center pushback could ultimately extend the cycle and reduce future supply by lowering capex and financing needs" - Morgan Stanley”

I trust nebius management and have little doubt that they will execute. Might dive into Bloom a little further. To me if bloom is truly legit, this is FUD and is the nebius buying opp I want to take advantage of. AI isn’t decelerating anytime soon imo unless there is a true bottleneck that squeezes it on the “supply” side

quote:

---

You copy and paste into the <> button

---

Thank you!

Zerohedge X post

Not sure how to embed the link so apologies there. But what are y’all’s thoughts on this? Trying to come up with reasons not to continue loading the boat at these prices.

(No Message)

He needs to go back and watch how we beat Lamar, Manziel, etc. Don’t over pursue, ends keep contain, and just make them beat you with the arm.

This one was from last year or something. Man some of these X wannabe’s piss me off. Blocking this guy

100%. The amount of alpha that retail investors have access to these days is unparalleled in history. Yes you have to filter out the bs, but there are diamonds in that rough worth looking for

Reading back through your post and the discord, it was him on the $35 sept calls hahaha

ST notes from 5/11:

[quote]ADEIA $ADEA VALUABLE SEMICONDUCTOR IP While Adea's main business is media IP, they have significant optionality in their semiconductor IP business, including hybrid bonding, advanced packaging, wafer-to-wafer and die-to-wafer technologies, advanced process-node technologies, microLED-related IP, and thermal-management technologies such as RapidCool. Their customers include Micron $MU, Sandisk $SNDK, SK Hynix, AMD, Samsung, Sony, UMC, and STM STMicroelectronics. Adeia’s semiconductor IP is valuable primarily because it sits at the intersection of advanced packaging, hybrid bonding, 3D integration, HBM, 3D NAND, chiplets, RF integration, image sensors, and AI compute thermals. This is not “chip design IP” like Arm CPU cores or Synopsys EDA/software IP. It is closer to process, packaging, interconnect & manufacturing-enablement IP. The highest-value part of the portfolio is hybrid bonding / DBI / DBI Ultra, not the broader generic “semiconductor" portfolio. That is where Adeia has the clearest evidence of blue-chip customer adoption. The industry problem is clear: Moore’s Law scaling is no longer enough. Chip performance is increasingly constrained by memory bandwidth, package interconnect density, thermal limits, power efficiency, and yield. Advanced packaging is becoming a primary scaling vector. Hybrid bonding directly addresses this. It gives semiconductor manufacturers a way to stack logic, memory, sensors, and RF devices with much denser electrical connections than traditional microbumps. Adeia says hybrid bonding is rapidly being designed into logic and memory products that support the AI ecosystem. SK HYNIX: In 2020, Xperi/Adeia announced a new patent and technology license agreement with SK hynix, including access to its broad semiconductor IP portfolio and a technology transfer of Invensas DBI Ultra 3D interconnect technology focused on next-generation memory. MICRON: In 2022, Xperi/Adeia announced a multi-year agreement with Micron under which Micron gained access to Adeia’s hybrid bonding IP to enhance next-generation memory devices. SANDISK/WESTERN DIGITAL: Adeia separately announced in 2023 that Western Digital entered into a long-term agreement to license Adeia’s semiconductor patent portfolio, including patents relating to hybrid bonding. Western Digital completed the separation of its Flash business in February 2025, and SanDisk became an independent public company under ticker SNDK. Adeia’s 2025 10-K then lists SanDisk among its semiconductor customers. SAMSUNG: Xperi announced in 2018 that it settled litigation with Samsung Electronics and entered into a new patent license agreement. AMD: Adeia’s lawsuits alleged AMD processors with 3D V-Cache infringed patents concerning hybrid bonding and advanced process nodes; AMD denied the allegations, and the case settled with AMD taking a multi-year license in March 2026. Adeia’s Q1 2026 release later specified that AMD licensed the semiconductor portfolio including hybrid bonding technology. INTERNAL INNOVATION ENGINE Adeia has approximately 13,750 patents and applications, including about 6,250 U.S. issued patents, 1,950 U.S. applications, 4,100 foreign issued patents & 1,450 foreign applications. Adeia says approximately 80% of its combined patent portfolio came from internal innovation. That is important. A pure patent aggregator can be more vulnerable to accusations that it merely buys and litigates. Adeia has a stronger argument that it is a long-term R&D organization that creates valuable technology. The company runs beautiful profitability metrics with 87% gross margins, 47% operating margins, and 27% net margins trading at 12x EV/EBITDA and 20x FCF.

Got in ELMT this morning at $18.64

Cerence $CRNC

In our robotics basket, one of our running thesis is the idea that automotive technology companies have cross-applicable platforms to a world of physical AI. Modern cars are robots. Those cars engaging with people and their environments has direct relevance to "robots" doing the same thing.

Our first two picks in this vein were $VLN @ $1.57 and $BB $6C @ $.87 both of those picks have worked *extremely well* so far.

The company currently trades at a very reasonable valuation (13 forward P/E, 6.5x FCF, 1.5x sales) with an explosive chart on all time frames (daily consolidating around 200sma, weekly consolidating around 50sma).

In 2025, Cerence technology was used in *52% of all automobiles produced worldwide*. Yes... you read that right, over half of every single car produced. When you interact with a car with your voice ("Hey Mercedes", "Hey BMW", "Hey Audi", "Hey Jeep", etc.) Cerence is most likely involved.

So, what's the AI angle? Voice AI is Cerence’s core market. The company’s historical competency is speech-driven control in noisy, embedded, real-world environments.

That matters because the car is one of the most difficult consumer environments for voice AI:

- background road/wind/cabin noise
- multiple passengers speaking from different seats
- intermittent connectivity
- safety-sensitive user commands
- multi-language/localization requirements
- OEM-specific cockpit APIs
- privacy and data-retention constraints

Many of these issues are cross-applicable to robotics. Cerence’s edge/cloud hybrid architecture is well-suited to this domain. Pure cloud AI can be more flexible, but cloud-only systems can fail when connectivity is poor or latency is unacceptable. Pure edge systems are more reliable and private but historically less capable. Cerence is trying to combine both.

The company is leaning into this cross-application. On the last earnings call:

*"Outside of automotive, consistent with what we have said in the past, we are concentrating our efforts on high-value verticals where our strength in edge solutions, quality, reliability, privacy, and a domain-focused approach matter most, and where we believe we have a clear right to win.

Specifically, we are prioritizing dealership AI, commercial and industrial operations, and select IoT and robotics applications. Rather than selling voice as a standalone component, our approach is to deliver full vertical solutions combining voice, LLMs and SLMs, orchestration, and workflow integration into purpose-built vertical packs."*

Cerence xUI is directly relevant here. Management describes xUI as a next-generation AI interface that can use Cerence’s CaLLM models, third-party LLMs, third-party agents, real-time data, and vehicle context. The company said in Q1 FY26 that it showcased next-generation LLM-powered experiences and new AI agents serving users across the vehicle purchase and ownership journey; in Q2 FY26, it said xUI-powered vehicles were entering production

Cerence’s capabilities — far-field audio, speech understanding, context-aware interaction, edge inference, cloud orchestration, wake-word, multi-zone awareness, and task execution — are portable to other embodied interfaces. The company already says its technology can apply beyond passenger cars to areas like two-wheelers, commercial trucks, TVs, smart watches, kiosks, and industrial automation markets. That means that the platform has clear cross-application potential for *physical devices* in general.

This name will *cross-apply* to our robotics + voice AI basket, but will be listed as Voice AI because that is the most relevant angle here for the current tech stack.

Filled $10C Aug 21 '26 @ $1.85 avg. 1.5% weighting.