strawberrywageslave

Engineer and scientist

[strawberrywageslave @ThUniqueOne]

This video covers generally about Caris Life Sciences $CAI, from their business operations to CEO's goal.

[strawberrywageslave @ThUniqueOne]

Lately I’ve been digging into Caris Life Sciences $CAI , an AI medical tech company focusing on precision oncology. Made a quick video to share. I’m currently holding some shares. If you're interested, I’m breaking down the company and the CEO’s story from a few different angles in next video.

[strawberrywageslave @ThUniqueOne]

🚀🚀🚀

[SpaceX @SpaceX]

Splashdown confirmed! Congratulations to the entire SpaceX team on an exciting tenth flight test of Starship!

[strawberrywageslave @ThUniqueOne]

@jdanielscarter The larger-than-expected drop in CPI may be interpreted as a sign of weakening overall demand and rising deflation risk.

[strawberrywageslave @ThUniqueOne]

The S&P 500 has begun its rapid decline. I will now start buying a little with each drop until my leverage reaches 1.5x.

[strawberrywageslave @ThUniqueOne]

I think the market is more resilient than I expected. I have already bought enough positions.

[strawberrywageslave @ThUniqueOne]

Current Market Expectations Regarding the U.S.-China Tariff War

From a stock market perspective, the current concern revolves around inflation triggered by the Trump administration’s reciprocal tariffs. However, regarding the state of the U.S. economy, after foreign investments in the U.S. surged in September last year, economic activity peaked in January this year and began to decline significantly starting in February.

As a result, foreign countries have gradually reduced their sell-offs of U.S. Treasury bonds, diminishing their need to stabilize their own currency exchange rates.

Currently, the buying interest in 10-year U.S. Treasuries stems from the expectation of fewer bond issuances in the future, coupled with concerns about the stock market’s lack of momentum from a high base. Interestingly, as more countries are willing to reach agreements with the Trump administration and increase investments in the U.S., they also seem more open to reducing tariffs across the board.

For emerging markets and developing countries, given their relatively low labor costs, broad tariff reductions seem to offer short-term economic recovery benefits. The biggest breakthrough appears to be the European Union’s willingness to lower tariffs to avoid a prolonged economic slowdown. The European Central Bank’s ongoing interest rate cuts also highlight the need for economic stimulus amid a sluggish economy.

Future Outlook: Digital Transformation and Economic Shifts

We may soon face circumstances vastly different from before, as various countries view investments in the U.S. as experimental bases for digital transformation in manufacturing industries. Investment projects from emerging markets in the U.S. are likely to focus on integrating AI and smart manufacturing as strategies to reduce corporate costs in the future.

Europe will face the most significant challenges, primarily balancing the costs of green energy with traditional energy sources to remain competitive amid intense global economic competition. European companies’ acceptance of digital transformation and AI-driven smart factories will heavily influence their competitiveness over the next decade.

U.S. companies are also under pressure to fully integrate AI into their operations, with 30% of companies already adopting AI to boost productivity and reduce labor costs.

Market Sentiment and Short-Term Projections

In the short term, most small and medium-sized enterprises are likely to face profitability pressures, due to persistently high interest rates and policy-driven transformation plans that will require a rapid increase in capital expenditure.

Currently, market sentiment is relatively pessimistic, reflecting a wait-and-see approach regarding the future growth potential of AI investments in businesses. Despite the S&P 500 reaching new highs—indicating investor confidence in the economy’s potential to accelerate again—it’s likely that the market will face a short-term decline over the next two weeks. This downward trend could continue until investors refocus on the upward revision of economic expectations driven by AI advancements.

[strawberrywageslave @ThUniqueOne]

Did you update the current 4o model because you plan to offer different computing power limits based on subscription tiers in the future, thereby increasing profitability and reducing the strain on computational support? @sama

[Sam Altman @sama]

we put out an update to chatgpt (4o).

it is pretty good.

it is soon going to get much better, team is cooking.

[strawberrywageslave @ThUniqueOne]

@elonmuskTN 67

[strawberrywageslave @ThUniqueOne]

The growing discussion around current quantum computers comes from Google's $GOOG significant breakthrough in quantum error correction. The importance of quantum error correction technology calls for revisiting the different types of quantum technology in order to understand how today’s commercial quantum computers will advance technology—and what opportunities we can discover along the way.

First, we need to address a question: Why do we need quantum computers? What are the differences between a general-purpose GPU and a QPU based on quantum computing? A GPU, which plays a critical role in deep learning and graphics rendering, excels at large-scale parallel data processing at room temperature, using classical bits for high-precision arithmetic. A QPU, however, operates at extremely low temperatures, using quantum superposition and entanglement (which offer an exponentially large state space) to solve problems where quantum advantage (exponential growth problems and nonlinear simulations) is significant.

Because quantum operations based on quantum logic gates (e.g., X, Z, CNOT) are easily affected by noise, quantum error correction is needed to increase single- or two-qubit gate fidelity. Note that once quantum computation is performed, the data-reading phase causes wavefunction collapse due to measurement, so measurement in quantum mechanics “fixes” a specific state. As a result, quantum computation results are probabilistic rather than high-precision numeric values, necessitating repeated runs (on the order of 1,000 times) to determine the distribution of outcomes.

Next we ask, What quantum technologies are available today? What are the distinctions among them? Which technology currently offers the most commercially scalable quantum computing solution? Currently, there are five main quantum technologies: Transmon, Ion Trap, Neutral Atoms, Photonics, and Spin Qubits. Before looking at the differences, advantages, and disadvantages of each, we need to understand three key metrics: “gate fidelity,” “coherence time,” and “operation speed.”

-Gate Fidelity:
In quantum computing, quantum logic gates are the basic units used to operate on qubits. An ideal quantum gate can precisely implement the intended quantum state transformations, but in real hardware, errors and noise cause deviations from these ideal results. “Gate fidelity” quantifies this deviation on a 0-to-1 scale, with values closer to 1 indicating higher accuracy. Ion trap and spin qubits perform best in fidelity, followed by transmon qubits, while neutral atoms and photonic qubits have comparatively lower two-qubit gate fidelities.

-Coherence Time:
Coherence time is a key physical property in quantum computing that describes how long a qubit can maintain its quantum state (superposition or entanglement). Coherence time dictates how many computational operations the qubit can undergo. A long coherence time enables more complex and deeper quantum circuits, whereas a short coherence time means higher error rates and an increased need for error correction.

Therefore, if coherence time is short, it requires higher gate fidelities as well as faster readout and computation. Thermal noise and electromagnetic interference disturb the qubit, causing gradual decoherence. This decoherence process disrupts the quantum state, making computations less accurate. Ion trap and photonic qubits have the longest coherence times, spin qubits and neutral atoms come next, and transmon qubits have relatively short coherence times.

-Operation Speed:
Operation speed is typically evaluated by the switching time of single-qubit gates. Different approaches to controlling qubits can significantly affect these time scales.

**Transmon Qubits are a type of superconducting qubit that use Josephson junctions to achieve nonlinear characteristics. They are controlled by adjusting DC flux or bias in a resonator (via Z controls to tune qubit energy levels, keeping their frequencies separate to avoid interference) and by microwave pulses (via X/Y controls for superposition states). Measurement of transmon qubits is done by detecting shifts in the resonator frequency (e.g., through reflection signals), with single-qubit gate times around 10–50 ns.

**Ion Trap Qubits suspend charged ions in a vacuum using electromagnetic fields, then use lasers to resonantly excite the ions, causing their spin states to superpose or flip. Single-qubit gates take around 10–20 μs.

**Neutral Atom Qubits use lasers to cool neutral atoms and trap them in optical lattices; microwave pulses then alter the electrons’ spin states. Single-qubit gates are on the order of 1–10 μs.

**Photonic Qubits encode quantum information in a photon’s polarization, path, or phase, using devices such as polarization rotators or wave plates to adjust a photon’s polarization or phase. Single-qubit gates range from 1 ps to 1 ns.

**Spin Qubits use magnetic fields or microwaves to control the states of electron spins or nuclear spins; single-qubit gates take about 10–20 μs.

In real-world applications and commercialization, the constraints of each quantum technology help us understand its development domain and direction:

-Photonic Qubits:
In theory, photonic qubits can have infinitely long coherence times, but practical limitations include optical losses and low gate-operation efficiency. Furthermore, current optical technologies are not easily integrated with classical control hardware, so photonic qubits are primarily used in quantum communication.

-Neutral Atom Qubits:
Because the operating environment is near room temperature without requiring ultra-low temperatures, neutral atoms are easily scaled for large arrays; however, they have lower gate fidelities and require stable optical fields. Commercial applications are not yet mature.

-Ion Trap Qubits:
They have long coherence times (on the order of seconds) and high two-qubit gate fidelities but suffer from slow operation speeds, limited scalability, and complex system architectures. Therefore, they are suitable for high-precision, long-duration operations but not for scenarios requiring high operation speeds.

-Spin Qubits:
With long coherence times, low energy consumption, small footprint, and potential for room-temperature operation, spin qubits offer many advantages. However, limited operation speed and lower two-qubit gate fidelity constrain their development, so current research focuses on quantum memory, quantum networks, or portable quantum computing devices.

-Transmon Qubits:
Built on mature semiconductor technology, they can leverage modern CMOS and superconducting processes and are well-suited for large-scale manufacturing. They also have fast operation speeds, flexible control, and good scalability, making transmon qubits today’s most commercially promising quantum computing technology. Companies such as Google, IBM, and Rigetti $RGTI use transmon qubits in their quantum computation units.

With Google introducing quantum error correction via surface codes (which expands physical qubits into fault-tolerant logical qubits), gate fidelity is greatly improved. As the technology behind quantum computing matures, research on room-temperature superconducting materials may narrow the scope needed to achieve superconducting capabilities, potentially lowering the cost of superconducting quantum computing.

The only major limitation of transmon qubits at present is that short coherence times limit computational depth. Qubits easily lose their quantum states during extended computation, requiring additional error correction procedures or algorithmic adjustments to reduce circuit depth. As a result, both error-correction capabilities and scalability of quantum computing units will command the greatest investment in today’s universal quantum computer landscape. Google and Rigetti $RGTI respectively lead the way in error-correction capabilities and scalability.

The challenge in quantum error correction lies in the no-cloning theorem (the impossibility of directly copying a quantum state) and the necessity to handle both bit-flip errors and phase-flip errors simultaneously. GOOGLE’s quantum error-correction technology, built on surface codes, arranges multiple neighboring physical qubits into one logical qubit, using redundant information for error detection and correction. By checking the state (stabilizers) of neighboring qubits for consistency, errors can be identified.

This approach divides physical qubits into “data qubits” and “measurement qubits,” with the latter detecting the relative state among neighboring data qubits; once an error is detected, a correction operation is triggered. As a result, surface codes require many physical qubits to form a single logical qubit. As the number of physical qubits increases, the logical qubit error rate decreases exponentially.

Regarding scalability, Rigetti adopts a modular architecture, launching the only multi-chip quantum processor currently capable of expansion. This means Rigetti can tackle even larger, more complex nonlinear problems, and by improving algorithmic operation time, there is a potential to combine both deeper computations and speed.

Our next question is: Given that GPUs and QPUs each have advantages and disadvantages, can their strengths be combined into a hybrid architecture, similar to how CPUs and GPUs coexist? In September 2022, Rigetti $RGTI and NVIDIA $NVDA formed a partnership in the quantum computing domain, focusing on CUDA Quantum (CUDA-Q) integration and co-developing workflows for hybrid GPU-QPU systems. Their collaboration involves quantum computing algorithm development, creating standard procedures for hybrid architectures in various fields, leveraging GPU compute power to simulate quantum hardware, and accelerating quantum error detection and processing. This deep collaboration will significantly speed up advancements and applications in transmon qubits.

For GPU-QPU applications, three areas stand out as having significant potential:

-Materials Development:
Designing materials and predicting their performance often require highly complex quantum-mechanical simulations, particularly for electronic structures and molecular dynamics. The complexity typically grows exponentially with system size, and cross-scale modeling involves nonlinear operations. QPUs have unique advantages in simulating many-body quantum interactions (e.g., optical and electrical properties of materials) and efficiently solving electronic structure problems (e.g., Hamiltonian solutions and ground-state energy calculations in quantum chemistry).

GPUs, meanwhile, handle large-scale parallel numeric simulations (e.g., forces among atoms and dynamic modeling) and macroscale modeling (e.g., heat conduction and elastic modulus). Current efforts include developing efficient catalysts, designing superconducting materials, and creating conductive polymers for wearable devices.

-Specific-Function Protein Synthesis & Protein Function Simulation:
This field involves computing high-dimensional molecular structures and interactions. Protein folding includes exponentially growing state spaces (energy-minimization problems), and protein functions depend on dynamic structural changes, requiring high-precision molecular dynamics simulations. QPUs are used to simulate the quantum electronic structure or local quantum interactions within proteins, harnessing VQE (Variational Quantum Eigensolver) to solve Hamiltonians in local regions of the protein.

GPUs handle large-scale classical molecular dynamics for the entire protein, including mechanical interactions among amino acids, e.g., with tools like GROMACS or AMBER. This synergy shortens the time to locate electronic distributions at active sites, determine a protein’s energy surface and folding pathways, and speeds drug development and gene engineering research.

-Artificial Intelligence/Machine Learning:
Use cases include seeking global optima in complex systems via quantum annealing (such as logistics route optimization and asset allocation) and building AI “consciousness,” viewed as a highly complex, nonlinear dynamic system. This involves massive data processing, nonlinear relationship modeling, and cross-dimensional information integration.

Simulating consciousness entails modeling perception (sensory input), memory (historical data), reasoning (logic models), and emotion (nonlinear features). Interactions among these layers are highly nonlinear and cross-dimensional. Hence, in multimodal emotion-based inference for virtual assistants, QPUs can provide high-dimensional nonlinear associative modeling, bringing AI closer to human cognitive abilities and emotional reasoning.

With these groundbreaking application examples, we can clearly recognize the enormous potential of GPU-QPU integration and gain deeper insight into the growth prospects of NVIDIA $NVDA, Google $GOOG, and Rigetti $RGTI.

[strawberrywageslave @ThUniqueOne]

The Federal Reserve’s expectations for rate cuts, inflation, and interest rate adjustments in 2025 reflect caution stemming from uncertainties surrounding Trump administration policies. While the Fed’s tone leans slightly hawkish, the overall U.S. economy is indeed performing well compared to the global economy.

Large inflows of capital into the U.S. have kept both the dollar and U.S. equities strong. Moving forward, the economic changes brought by manufacturing reshoring and policies targeting the expulsion of illegal immigrants will be key areas of focus for Federal Reserve officials.

[strawberrywageslave @ThUniqueOne]

Today, I want to discuss the second-tier leaders in two different fields: the AI server chip designer $AMD and the health insurance provider $ELV, focusing on the opportunities and potential growth they present.

*The current demand and growth of AI servers are well-known and widely observed. While $NVDA dominates with a significant market share, it’s easy to overlook that the overall AI server market demand is also continually being revised upward. This means that during this high-growth phase, companies investing in this space can continue to earn substantial returns.

Analyzing AMD’s business reveals that its Data Center and Client segments are growing rapidly, while Gaming and Embedded segments are shrinking in contribution. The combined revenue share of Data Center and Client increased from 74% last quarter to 80%, indicating that these two segments will be the primary drivers of revenue growth. Based on AMD’s Q3'24 guidance for the next quarter, we can expect its price-to-earnings (PE) ratio to approach the multiple currently attributed to NVDA.

Given the solid landing of AI applications, AMD’s value is approaching a reasonable price compared to current market benchmarks. If the next earnings report does not result in a positive market reaction, $AMD would become an undervalued stock. However, this seems unlikely. With the foreseeable recovery of the consumer market, it’s highly probable that AMD’s outlook will be revised upward, making now a good time to build a position in AMD.

*In the health insurance sector, Elevance Health’s Q3'24 earnings revealed the reasons for its significant drop in free cash flow (FCF):

-Decline in Membership: As of September 30, 2024, ELV’s total medical membership was approximately 45.8 million, down by 1.5 million compared to the same period last year. This was primarily due to Medicaid members losing coverage as a result of eligibility redeterminations.

-Rising Medical Costs: Increased medical needs among Medicaid members drove up medical costs, further impacting the profitability of this segment.

-Mismatch Between Rates and Costs: Many states use outdated data to set Medicaid reimbursement rates, which do not reflect current cost increases, thereby widening the gap between revenues and expenditures.

These issues have caused a decline in earnings per share (EPS), raising investor concerns about the high medical costs and low returns of the Medicaid business. This reflects the significant time lag between government pricing and actual medical demand in public health programs. However, some fundamentals remain unchanged, including a year-over-year increase of 600,000 commercial members and total quarterly revenue of $44.7 billion, representing over 5% annual growth.

Thanks to Carelon’s AI-based digital health services platform, $ELV has maintained overall revenue growth, signaling that the company is at a critical juncture of digital transformation and product diversification. Drawing insights from UnitedHealth Group $UNH, we see that AI-driven digital health services platforms have contributed to excess returns in revenue and significantly reduced cost volatility stemming from government healthcare subsidy policies. Furthermore, the issues currently disclosed can be classified as short-term challenges, with the CEO indicating during the earnings call Q&A that these impacts will diminish next quarter.

Considering all these factors, the second-tier leader’s current stock price appears attractive, presenting a potential investment opportunity.

[strawberrywageslave @ThUniqueOne]

Today, I want to discuss a company with potential for rapid growth, CyberArk $CYBR, which might benefit from riding on Cisco's $CSCO coattails and begin to shine in the second half of 2025.

CyberArk is a platform provider focused on identity security, ensuring secure access to an organization's critical business data and infrastructure, preventing data breaches caused by identity and credential misuse. The platform employs intelligent privilege controls to seamlessly protect various workloads in hybrid and multi-cloud environments, including multi-identity management, privileged access management (PAM), endpoint privilege security, and machine identity management. Currently, its revenue comes from three main sources: Subscription, Perpetual license, and Maintenance and professional services, which account for 73.1%, 1.1%, and 25.6% of total revenue, respectively. Currently, 47% of ARR comes from the banking/financial, manufacturing, and government sectors, with a total ARR growth rate of 31% year-over-year.

Here are the detailed revenues of the three major business lines:
*Subscription: $175.57M, +42.8% YoY
*Perpetual license: $2.8M, -28.6% YoY
*Professional Service: $61.6M, -4.16% YoY
*Total Revenue: $240.1M, +25.5% YoY

Revenue by geographic region:
*US: 59%
*EMEA: 31%
*APJ: 11%

Costs and Expenses:
*Cost of Revenue: $47.18M, +13% YoY
*Subscription: $24.56M, +15% YoY
*Perpetual license: $0.46M, -27% YoY
*Professional Service: $22.15M, +11.8% YoY

*Total Operating Expense: $204M, +16.4% YoY
*Sales and Marketing (S&M): $113.69M, +15% YoY
*General and Administrative (G&A): $31M, +25.8% YoY
*Research and Development (R&D): $59.3M, +14.6% YoY

Income and Profit:
*Operating Loss: $11M, -56.8% YoY
*Net Income: $11.1M, +104.85% YoY
*Free Cash Flow: $52M, +271% YoY

One key highlight is the strategic partnership between Cisco and CyberArk, aiming to develop the CyberArk Identity Security Platform and introduce CyberArk Dynamic Privileged Access into Cisco. This will significantly reduce the attack surface by providing real-time access privileges to thousands of administrative users. The partnership includes identity security platform integration, multi-factor authentication collaboration, automated identity governance, security enhancement for DevOps environments, and cloud infrastructure protection.

This partnership means that CyberArk, targeting the high-value market, will be rapidly adopted in the IoT sector through Cisco's comprehensive software and hardware solutions, expanding its market from large enterprises to small and medium businesses. This gives CyberArk greater growth potential, and its future development is worth watching.

[strawberrywageslave @ThUniqueOne]

Tonight, South Korea announced martial law. Without discussing how long it will last, we are likely to see the suspension of the entertainment industry and a slowdown in economic activities. This could have a significant impact on the current semiconductor industry in South Korea, and the server industry, which needs to maintain supply, will likely turn to second sources for assembling servers. If martial law is prolonged, it will exacerbate currency devaluation and rising inflation, and the liquidity of the real estate sector will also be greatly affected. Hopefully, the situation does not worsen further.

[strawberrywageslave @ThUniqueOne]

When discussing the process of digital transformation, most companies are gradually realizing that with the acceleration of AI and the rise of remote work, it is inevitably a necessary capital expenditure. Enhancements in edge computing, IoT, enterprise networks, and AI will significantly improve operational management, work, and production efficiency, while also driving the demand for enterprise network security management. However, only a few companies are in a favorable position to provide comprehensive solutions across multiple fields, and Cisco $CSCO is one of them. In the following sections, we will discuss the key areas of this company.

Cisco is a global leader in networking technology, offering solutions that encompass networking hardware, software, and security. As one of the few companies providing comprehensive solutions, Cisco's future development focuses on two areas:

*Integrated Hardware and Software Solutions for Enterprise Networks:
With the growing prevalence of remote work and hybrid work models, the demand for secure and efficient network connectivity is increasing. Cisco's network solutions include hardware (such as routers and switches) and software services (such as network management and collaboration tools), helping enterprises reduce complexity and avoid compatibility issues between systems and applications, thereby significantly reducing the time cost of transformation.

*Efficient Enterprise Operations Solutions Combining IoT, Edge Computing, and AI:
With the development of IoT and AI integration, concepts like smart factories and smart cities have emerged to help enterprises and governments improve operational efficiency. They also enable data that was previously unquantifiable to be recorded and analyzed, providing concrete solutions to existing problems. The introduction of edge computing reduces the dependence on centralized computing, addressing latency issues and enhancing security in production management monitoring.

To achieve this, Cisco acquired Splunk, Robust Intelligence, and DeepFactor in 2023 and 2024.

"Splunk focuses on data analysis and big data management solutions for collecting, searching, monitoring, and analyzing massive machine data generated by various applications, systems, devices, and IT infrastructure. Splunk is proficient in handling both structured and unstructured data, with unique advantages in real-time data analysis and monitoring. Its products, including Splunk Enterprise, Splunk Cloud, Splunk IT Service Intelligence (ITSI), and Splunk Security Information and Event Management (SIEM), have been integrated into Cisco's network and security solutions for real-time detection and response to network threats, providing more comprehensive end-to-end IT security and operational solutions.

Robust Intelligence specializes in AI security solutions, helping enterprises reduce risks and errors when deploying AI applications, enhancing product security in the AI field, and ensuring AI models can operate in demanding ecosystems.

DeepFactor focuses on the security of cloud-native applications, helping developers identify potential vulnerabilities and risks in development and runtime environments, providing stronger application security solutions for modern enterprises in multi-cloud and hybrid environments."

Since acquiring Splunk, the number of product orders has increased by 20%, with 11% of that growth attributed to Splunk. It is expected that the revenue of the other two companies will gradually be integrated from Q2 onwards. With the current technologies combined, a significant growth curve is expected to emerge after two quarters, which will also include normalized traditional order demand. Financial information worth noting includes three aspects:

*Security growth of 100% and Observability growth of 36% have brought billion-level growth in a short period. This indicates the rapidly increasing demand for Network Security, Secure Access Service Edge (SASE), Threat Intelligence, network monitoring, and analytics, aligning with our previous descriptions. In an accelerated AI environment, enterprises' need for cybersecurity and analytics is becoming increasingly evident and will likely be the first wave of AI application.

*RPO (Remaining Performance Obligations) reached $40 billion, growing by 15%, with 49% of contracts to be recognized after one year, indicating visible demand sustainability. As consolidation continues, we may see higher RPO growth in the future.

*Capital expenditures increased by 61% compared to the same period last year, while labor expenses were reduced this year, suggesting the organization is moving forward toward future growth directions, focusing on long-term development. Furthermore, a slightly lower EV/EBITDA indicates that the market may underestimate the company's profitability after consolidation.

However, regarding competitors, particular attention should be paid. Huawei also offers comprehensive solutions, and Cisco faces significant competition from Huawei's lower prices. The impact of political support is substantial and uncontrollable, which introduces a lot of uncertainty that needs to be factored in to lower price expectations and thus reduce risk. Considering the above observations and analyses, a suitable time for increasing positions would be during small-cycle corrections in the coming months. However, attention should be paid to competitor-related news to assess the upper limit of positions for risk control purposes.

[strawberrywageslave @ThUniqueOne]

In recent days, I have been seriously thinking about the latter part of the bull market, particularly what sectors and companies I should invest in as we may face a future correction. My first idea is the recovering semiconductor sector, which is gradually recovering by the end of this year after experiencing inventory adjustments and weak end-market demand. During a future market correction, semiconductor-related companies may see valuations approach reasonable levels.

My second idea is to invest in companies with comprehensive communication semiconductor solutions. Based on the 10-year cycle of communication technology advancements, the 5G technology introduced in 2019 is entering its latter stage, meaning the phase of gradual popularization. This phase will significantly increase sales of 5G base stations and mobile devices, which will also provide some support to stock prices during a downturn.

The third idea is companies with software solutions supporting digital transformation. With the current growth in AI, the use of data for optimization, decision-making, and automation is an inelastic demand. Even in times of economic downturn caused by certain events, the market for these solutions will continue to expand, although investment levels will be lower than during economic prosperity. Companies that meet these criteria can enhance confidence in left-side trading and are suitable for monitoring during the irrational prosperity stage of the bull market. Fortunately, I have found a company that meets these conditions: Broadcom $AVGO.

I believe many are familiar with this company, but not necessarily with the details of its business. Broadcom is a leading global provider of semiconductor and infrastructure software solutions, covering a wide range of sectors. Its primary revenue sources are divided into two parts: semiconductor solutions and infrastructure software. Semiconductor solutions are further divided into five segments: Networking, Wireless, Broadband, Server Storage, and Industrial. Infrastructure software offers a series of software solutions for enterprise IT environments, including mainframe software, automation and DevOps tools, enterprise security solutions, network performance management, storage management, AIOps, and API management. These products help enterprises manage their complex IT infrastructure and achieve digital transformation goals, simplify IT operations, improve efficiency, and ensure system security and availability.

In the infrastructure software business, a notable focus is VMware, acquired in 2022 for $61 billion. VMware is a provider of virtualization and cloud infrastructure solutions, helping enterprises build and manage modern IT infrastructure with virtualization, cloud management, hybrid cloud, and multi-cloud environments. VMware's core businesses include compute virtualization (vSphere), storage virtualization (vSAN), network virtualization and security (NSX), cloud management and multi-cloud platform (vRealize Suite and VMware Cloud Foundation), application modernization and Kubernetes platform (Tanzu), and desktop and application virtualization (Horizon).

These solutions enable enterprises to achieve efficient IT infrastructure operations and management and gradually expand into application modernization and multi-cloud management, meeting diverse needs in digital transformation. Following Broadcom's acquisition of VMware, they canceled perpetual licenses, fully transitioned to subscription licensing, and adopted a consumption-based pricing model to enhance resource flexibility for small and medium-sized enterprises while reducing initial investment pressures. To further support rapid deployment, testing, and continuous delivery of enterprise applications, VMware enhanced its container management capabilities (Kubernetes) by acquiring Heptio in 2018 and subsequently launched VMware Tanzu, which has been integrated into VMware Cloud Foundation.

"A container is a lightweight virtualization technology that packages an application and all its dependencies (including libraries, configuration files, etc.) together to ensure it can run consistently in any container-supported environment, effectively achieving 'develop once, run anywhere.' Containers are managed by container engines like Docker and orchestration tools like Kubernetes. Containers do not require a full operating system; they share the host OS kernel and use isolation technologies like cgroups and namespaces to achieve relative isolation of applications.

As a result, containers start very quickly, typically within seconds. This technology significantly improves resource utilization and application deployment flexibility by reducing system redundancy and enhancing application portability. It is particularly suitable for microservices architecture and DevOps applications. VMware Tanzu is VMware's container management and Kubernetes platform that helps enterprises deploy, run, and manage containerized applications. Additionally, VMware has optimized and lightweighted virtual machines through designs like Thin Provisioning, resource reservations and limits, using Photon OS, lightweight VM templates, and VMware Horizon for desktop virtualization, helping small and medium-sized enterprises effectively run multiple VMs with limited resources, reducing resource occupation, and improving resource utilization."

Currently, VMware represents 65% of software revenue within Broadcom's infrastructure software business, and the overall software segment contributes 44% of total revenue, marking a 47% year-over-year growth. This trend bodes well for pushing EBITDA beyond $8.5 billion in the next fiscal year. VMware Cloud Foundation plans to release updated versions in March and July 2025, meaning that this integrated solution platform for compute, storage, network, and management functions will be fully available mid-next year, including extensions for integrating public and private clouds, which is expected to drive greater adoption among small and medium-sized enterprises, offering increased flexibility for resource allocation based on business needs. However, compared to other competitors in the Kubernetes open-source ecosystem (such as Red Hat OpenShift), there is still work to be done. Broadcom is expected to further expand developer support and enhance integration with open-source technologies, particularly in the Kubernetes and DevOps fields, to build a complete developer ecosystem and invest continuously in developer tools and community building.

Moving on to Broadcom's semiconductor business: hyper-scale customers are continually expanding their AI clusters, with custom AI accelerators growing by 3.5 times year-over-year. For Networking, revenues are expected to increase by 40% in Q4. Broadband and Industrial segments have bottomed out and are now on the path to recovery. The Server Storage business is growing quarterly but may take a few more quarters to achieve revenue breakthroughs. Wireless revenues are expected to remain relatively flat year-over-year with the introduction of next-generation devices and North American customers. Overall, the non-AI market has already hit its low, and recovery is expected in Q4.

Particular attention should be given to the low Earth orbit satellite communication solution, which includes high-performance RF front-end solutions (RF transceivers, FBAR filters, and power amplifiers), mixed-signal ASIC, network processors for satellite data communications, and optical communication modules. In times of geopolitical tension, drawing lessons from the ongoing Russia-Ukraine war, many countries are accelerating the establishment of secure communication networks. And that has garnered significant attention for companies like $ASTS, $RKLB, and SpaceX. Broadcom has comprehensive solutions in the LEO satellite communication technology domain, and its high integration and low power consumption design meet the stringent requirements of LEO satellites. Satellite communications are likely to become a growth driver for Broadcom, with Qualcomm as a potential competitor.

Based on the above assessments, I plan to continue buying Broadcom shares during significant price drops, with this being my goal for the next five months. Considering the potential price volatility during the irrational prosperity phase, part of my AVGO shares will serve as swing trade positions.

[strawberrywageslave @ThUniqueOne]

Every time $NVDA announces its earnings report, its stock price becomes “dopamine-driven.” As long as the growth trend remains unchanged, continuing to hold and increase your position is the best way to achieve optimal returns.