The memory aspect of semiconductors today has gotten so extreme. Stuff is so expensive that people are simply not able to make lower-end equipment or like devices anymore. And this is like killing everything, right?
AI chips make like 65% operating margins and gaming does like 40%. So obviously from a business perspective it doesn't really make sense to put too much effort into GPUs which is kind of sad you know because what happened to the rest of us you know everything is like AI.
Meta's platform of apps has 3.5 billion daily active users, and they make something like I think it's like $200 a year off of each user in advertising, which just goes to show that like for every person in the world, there's a lot of companies that want to sell them something.
The AI era is fundamentally reorienting the semiconductor industry from consumer-driven volume to enterprise-driven performance and specialized memory. This means sustained, massive capital expenditure from hyperscalers will continue to be the primary growth engine.
Invest in companies providing specialized memory (HBM, high-density NAND) and custom silicon solutions for AI workloads. These components are the bottlenecks and profit centers for hyperscalers.
The AI infrastructure buildout is far from over. Expect continued, accelerating investment in compute and memory through 2027 and beyond, creating a "rising tide" for the entire semiconductor supply chain.
AI's insatiable demand for compute and memory is fundamentally re-prioritizing semiconductor manufacturing, shifting capacity and R&D from consumer products to high-margin data center components. This creates a new economic reality where memory is the bottleneck and a strategic asset.
Invest in companies positioned to supply high-performance memory (HBM, advanced DRAM, NAND) or those hyperscalers with clear, high-margin internal monetization paths for their AI capex (e.g., advertising-driven models).
The AI infrastructure buildout is far from over, with hyperscalers projecting continued, accelerating capex into 2027 and beyond. This sustained investment will keep memory prices elevated and drive innovation in optical interconnects and custom silicon, creating both challenges for consumers and immense opportunities for strategic investors and builders.
AI's pervasive influence is fundamentally re-architecting the semiconductor supply chain, shifting investment from consumer-grade components to high-margin, specialized AI memory and compute, creating a sustained demand cycle.
Invest in companies positioned to capitalize on the broad memory demand, from HBM manufacturers to NAND suppliers, and those hyperscalers with clear, high-margin monetization paths for their AI infrastructure.
The AI infrastructure buildout is far from over, with hyperscalers committing hundreds of billions annually. This sustained investment will continue to drive semiconductor prices and innovation, making memory and specialized compute the critical bottlenecks and opportunities for the next 3-5 years.
Skyrocketing Costs: GDDR7 prices have quadrupled in the last year, with DRAM contract prices doubling in a single quarter. This means the memory (VRAM) now accounts for 80% of a gaming GPU's bill of materials, making consumer GPU manufacturing increasingly unprofitable.
AI's Profitability: AI chips offer significantly higher operating margins (65%) compared to gaming GPUs (40%). This incentivizes companies like NVIDIA to focus on data center AI, meaning less investment in consumer products and a clear business rationale for the current market dynamics.
Enterprise Skepticism: Wall Street is wary of Microsoft's AI capex due to longer enterprise sales cycles and less immediate ROI compared to advertising-driven models. This suggests investors are prioritizing quick, high-margin returns in the current AI gold rush.
The memory aspect of semiconductors today has gotten so extreme. Stuff is so expensive that people are simply not able to make lower-end equipment or like devices anymore. And this is like killing everything, right?
Capex Surge: Google, Meta, Amazon, and Microsoft are collectively committing over $600 billion in capex for 2026, a 70% average increase. This massive investment is primarily directed at building out AI data centers, compute, memory, and networking infrastructure.
NAND's Moment: Nvidia's Vera Rubin platform will feature over 1,152 terabytes of NAND per rack, with Morgan Stanley estimating Reuben alone will consume 13% of global NAND supply by 2027. This highlights the critical role of massive, cheaper storage for context memory and KV cache in scaling AI.
The memory aspect of semiconductors today has gotten so extreme. Stuff is so expensive that people are simply not able to make lower-end equipment or like devices anymore. And this is like killing everything, right?
We're in an era of finding a use case for something that just requires so much memory. This I I don't see it changing in the immediate future.
AI chips make like 65% operating margins and gaming does like 40%.
AI's integration into core business models is driving hyperscalers to commit unprecedented capital to infrastructure, shifting semiconductor demand from consumer-driven cycles to enterprise-grade, high-margin AI components.
Investigate memory manufacturers and specialized AI silicon providers, as their products are becoming the foundational bottleneck and highest-margin components in the AI infrastructure buildout.
The AI capex spend, projected to exceed $600 billion in upcoming years, is a rising tide lifting all semiconductor boats. Understanding where this capital flows—from HBM to NAND and custom silicon—is crucial for positioning your portfolio and product roadmap for the next half-decade.
AI's computational hunger is fundamentally re-architecting the semiconductor industry, shifting focus from consumer-driven volume to high-margin, specialized memory and compute for hyperscalers. This means a sustained, elevated demand for advanced silicon, with traditional consumer markets becoming a secondary concern.
Invest in companies providing core AI infrastructure components—HBM, advanced NAND, and custom silicon design capabilities—or those hyperscalers with clear, high-margin monetization paths for AI, like advertising.
The AI infrastructure buildout is far from over, with hyperscalers projecting continued, accelerating capex into 2027 and beyond. This sustained investment will keep memory prices high and demand for specialized AI hardware robust, creating a new economic reality for tech investors and builders.
The US is pivoting from a QE-fueled, government-led economy to a "free market" model under the new Fed Chair, Kevin Warsh. This means a potential reduction in the Fed's balance sheet (QT) and lower rates without yield curve control (YCC), leading to decreased US dollar liquidity.
Adopt a phased, data-driven allocation strategy. Michael Nato recommends an 80% cash position, deploying first into Bitcoin (65% target) at macro lows (around 65K-58K BTC, MVRV < 1, 200WMA touch), then into high-conviction core assets (20%), long-term holds (10%), and finally "hot sauce" (5%) during wealth creation.
The current "wealth destruction" phase, while painful, presents a rare opportunity to accumulate assets at generational lows, provided one understands the macro shifts and adheres to a disciplined, multi-stage deployment plan.
The financial world is splitting into two parallel systems: opaque TradFi and transparent onchain finance. Value is migrating to platforms that can simplify and distribute onchain financial products globally.
Invest in or build applications that prioritize mobile-native experiences, abstract away crypto complexities (like gas fees), and offer tangible real-world utility for onchain assets.
The future of finance is onchain, and "super apps" like Jupiter are building the necessary infrastructure and user experiences to onboard the next billion users.
Crypto's initial broad vision has narrowed to specific financial use cases, while AI and traditional markets capture broader attention. This means builders must focus on tangible value and investors on proven models.
Identify projects with novel token distribution models (like Cap's stablecoin airdrop) or those building consumer-friendly applications within new ecosystems (like Mega ETH) that address past tokenomics failures.
The industry is past its naive, speculative phase. Success hinges on practical applications, robust tokenomics, and competing with traditional finance, not just abstract ideals.
The Macro Shift: From unbridled, community-driven idealism to a pragmatic, business-focused approach. Early crypto imagined a world where "everything is a thing on Ethereum," but reality has narrowed its primary use cases to finance and trading, forcing a re-evaluation of tokenomics and community models. This shift is also driven by AI capturing mindshare and traditional finance co-opting blockchain tech.
The Tactical Edge: Re-evaluate token distribution models. Instead of relying on inflationary yield farming that creates sell pressure, explore innovative approaches like Cap's "stable drop" (airdropping stablecoins, then inviting participation in a token sale) to align incentives and attract long-term holders. Focus on building real products with defensible business models, even if they lean more "business" than "protocol."
The shift from centralized, static data aggregation to decentralized, real-time, incentivized intelligence networks is fundamentally changing how data-intensive industries operate.
Investigate subnet opportunities where incumbent data quality is low and validation is a core challenge.
The future of sales is not just about more leads, but smarter, fresher, and more relevant ones.
The Macro Shift: As trust erodes in traditional financial systems and geopolitical risks rise, capital is flowing towards more efficient, permissionless DeFi markets. This is forcing traditional finance to adapt or lose market share.
The Tactical Edge: Evaluate DATs trading below NAV for potential M&A or activist plays, as these discounts often reflect management misalignment rather than fundamental asset weakness.
The Bottom Line: The current market volatility, Fed policy shifts, and the rise of DeFi are not just noise; they are reshaping capital allocation. Investors and builders must understand these structural changes to position for the next cycle of institutional adoption.
