Chip Stock Investor Podcast

Amazon, Microsoft, Alphabet, and Meta just collectively committed to over $750 billion in capital expenditure for 2026. They spent $130 billion in a single quarter. That is a 70% increase from what these companies spent in 2025 — and the spending is still accelerating into the second half of the year.

The ROI is showing up. Operating margins are expanding across all four businesses. Google Cloud grew 63% year over year. AWS grew 28%. Microsoft Intelligent Cloud grew 29%. Meta grew revenue 33%. This is not speculative infrastructure spending anymore. These are some of the most profitable businesses ever built, getting more profitable.

But the more important conversation is about what this means for everyone else.

CSI has a framework for understanding how AI infrastructure investment actually flows — and it is the most useful mental model for investors trying to figure out where value accrues in the AI buildout. The hyperscalers eat first. They buy the technology and deploy it internally before any customer touches it. Their strategic investment partners eat second — OpenAI, Anthropic, and others who receive capital and get early infrastructure access. Enterprise software companies and Neo Cloud providers eat third. They get the leftovers, and right now they are scrambling.

This creates two distinct problems. Neo Cloud companies have great infrastructure but no vertical integration — no final product of their own. The moment spare capacity appears in the market, their economics break down rapidly. Enterprise SaaS companies have great products but no infrastructure control — they get stuck waiting for technology that the hyperscalers have already been using internally for years.

CSI lays out what both camps need to do to survive the next phase: Neo Clouds need to start developing software and services of their own before excess capacity forces their hand. Enterprise software companies need to start acquiring infrastructure assets — and there are already early signals that the smarter ones are doing exactly that.

This episode was released to Semi Insider members several weeks before this public version. Members receive CSI's full research, live Q&A sessions, and analysis like this as it happens — not weeks later. If that matters to you, the membership page is at chipstockinvestor.com

What we cover:
— Why hyperscaler earnings reactions are about cashflow expectations not beats or misses
— $750B+ in 2026 AI CapEx — full breakdown across Amazon, Microsoft, Alphabet, and Meta
— Amazon Q1 2026: AWS +28% YoY, operating margin expanding to 13.1%
— Microsoft Q1 2026: Intelligent Cloud +29% YoY, 46.3% operating margin
— Alphabet Q1 2026: Google Cloud +63% YoY, 36.1% operating margin
— Meta Q1 2026: +33% revenue, CapEx raised to $245B, why the stock reaction was muted
— The ROI is real — operating leverage across all four hyperscalers
— The "who eats first" hierarchy — the most useful AI investing framework right now
— Neo Cloud companies — the vertical integration problem and what needs to change
— Enterprise SaaS — why they are chasing the puck and what the smart ones are doing
— Early signals: Salesforce, Snowflake, Fortinet, Trade Desk CapEx moves

Disclosure: Nick and Kasey hold positions in several companies mentioned. This content is for general information only and is not individual investment advice. All investing involves risk.

chipstockinvestor.com

In 2000, venture capitalists funded 200 semiconductor companies. By 2015, that number had fallen to single digits. The cost of designing a chip went from $33 million in the late 1990s to over $200 million today, and it takes four or more years to complete. There are not enough engineers. The timelines are too long. The investment risk is too high.

Faraj Aalaei has spent three decades watching this problem build. He co-founded Centillium Communications and took it public on Nasdaq in 1997. He co-founded Aquantia, took it public in 2017, and sold it to Marvell Semiconductors in 2019. He then spent several years investing in AI companies before recognizing that generative AI could do for chip design what it has done for software — compress timelines, reduce cost, and open the field to a much wider group of contributors.

In 2024, he founded Cognichip with a straightforward but ambitious vision: everybody should be able to be a chip designer.

Nick sits down with Faraj for a wide-ranging conversation that covers the full landscape — why the semiconductor industry has a structural problem that EDA companies like Synopsys and Cadence cannot solve alone, why generic large language models fall flat when applied to chip design, how Cognichip trains physics-informed models on proprietary synthetic data without touching any customer IP, and what actually happens to the semiconductor industry when design timelines collapse from years to months.

The implications are significant. Hyperscalers like Google, Meta, and Microsoft could design more bespoke ASICs faster and cheaper. Foundries like TSMC would see shorter cycles between customer signup and first wafer revenue. Startups that today cannot raise funding because the design cost is too high would suddenly become investable. And companies like Nvidia — which already iterate faster than anyone in the industry — could move faster still.

The conversation also covers Cognichip's differentiation from agentic workflow competitors, the fundamental reason why physics-informed models are necessary for semiconductor design in a way they are not for software, and — at the very end — the IPO question.

What we cover:
— Why chip design went from $33M to $200M+ and VC funding collapsed
— The engineer shortage and why semiconductors lose to software for talent
— Why the 6-year chip design lag is a fundamental problem for AI progress
— EDA companies and AI — complementary rather than competitive
— How Cognichip trains models without using customer IP
— Physics-informed AI vs. generic LLMs — why the distinction matters
— The vision: anyone can design a chip — what that actually means
— What happens when design timelines collapse — impact on Nvidia, startups, foundries
— Cognichip vs. agentic workflow competitors — the fundamental model difference
— Hyperscaler ASIC strategies and CapEx implications
— Manufacturing yield improvement and AI's role
— The IPO question

Disclosure: Cognichip is a private company and is not publicly investable at this time. This content is for general information only and is not individual investment advice.

chipstockinvestor.com

Over a year ago, CSI did a three-part deep dive on co-packaged optics after Nvidia dedicated an entire segment of its GTC keynote to the technology — naming Lumentum and Coherent as the primary beneficiaries. The analysis was right. They did not buy.

That mistake is now worth talking about directly.

Lumentum just reported fiscal Q3 2026 revenue up 90% year over year. Q4 guidance implies triple-digit year-over-year growth. Nvidia made a $2 billion investment in both Lumentum and Coherent, and separately announced a major fiber optic cable manufacturing expansion with Corning. Co-packaged optics products have not even begun shipping in volume yet — that catalyst hits in December 2026. The case for Lumentum continues to build.

But this is CSI, and true conviction in a business means covering what could go wrong as well as what is going right. There is a significant dilution story unfolding that every Lumentum shareholder needs to understand before adding to a position.

When the stock was trading at roughly one-tenth of its current price, Lumentum raised cash by issuing convertible notes — a type of debt that converts to equity when the stock reaches certain price milestones. The stock has now blown through those milestones. All of that convertible debt is now eligible to convert into stock at terms that are extremely favorable for the debt holders and extremely expensive for existing shareholders. The result: shares outstanding are expected to increase by approximately 20% over the next two quarters. Nick and Kasey explain the full mechanics clearly — why it happened, what it costs, and whether the revenue acceleration can outrun the dilution.

Also covered: the Qorvo fab acquisition in North Carolina that adds indium phosphide manufacturing capacity in two to three years, and what operating leverage looks like when a company goes from negative margins to all-time highs in the span of a few quarters.

What we cover:
— Why CSI did the deep dive on co-packaged optics and still did not buy — the honest lesson
— Lumentum fiscal Q3 2026: 90% revenue growth — what drove it and what comes next
— Q4 guidance: triple-digit YoY growth before CPO products even ramp
— Nvidia's $2B investment in Lumentum and Coherent — the supply chain signal
— Nvidia and Corning fiber optic expansion — Nvidia's hands all over the supply chain
— Co-packaged optics — the December 2026 catalyst that has not landed yet
— Operating leverage in action: from negative margins to all-time highs
— Convertible notes explained: why ~20% share dilution is coming in 2026
— Qorvo North Carolina fab acquisition — InP capacity coming in two to three years
— The bottleneck in laser module manufacturing and why Lumentum dominates it

Sponsored by fiscal.ai — 25% off any paid plan through May 14 only. Use our link: fiscal.ai/csi

Disclosure: Nick and Kasey hold positions in Lumentum and Coherent. This content is for general information only and is not individual investment advice. All investing involves risk.

chipstockinvestor.com

Intel stock is up big over the last year. On the stock price, CSI was wrong — and they are saying so directly.

On the business analysis, they are standing firm.

AMD is steadily taking data center CPU market share from Intel. The driver is availability — both companies rely on TSMC for their most advanced chiplets, but AMD has used that availability more effectively in a CPU shortage environment where demand is outpacing supply. After AMD's most recent earnings, the trajectory is clear: if things continue, AMD could pass Intel in data center and AI revenue as early as late 2026 or sometime in 2027.

Intel's client segment — the portion of the business keeping the lights on — continues to face market share pressure from AMD. The turnaround story is real and the early innings are genuinely encouraging. But Intel at 104x forward earnings needs everything to go right, and a lot still needs to go right. AMD at 48x, with cleaner data center momentum and a more consistent growth trajectory, remains the cleaner pick — even after being wrong on Intel's stock price.

This episode also covers two important supporting stories. Lattice Semiconductor is quietly back — up 42% in its most recent quarter, with record revenues possible as early as Q2 2026, and an AMI software acquisition adding roughly $1 billion in annualized revenue by year end. And AMD's embedded FPGA segment, inherited from the Xilinx acquisition, remains a highly profitable cushion even through its current growth trough.

The close leaves listeners with one more thing to watch: the Vera CPU, coming later in 2026.

What we cover:
— Why CSI was wrong on Intel's stock price — and why the business analysis still holds
— AMD vs. Intel data center CPU market share — the trajectory and what drives it
— The CPU shortage and AMD's TSMC availability advantage
— Intel client segment — profitable, but losing ground quarter by quarter
— Intel at 104x forward P/E vs. AMD at 48x — what each multiple actually implies
— Lattice Semiconductor: +42% quarterly, record revenues in sight for Q2 2026
— Lattice AMI acquisition — $1B annualized revenue run rate by year end
— AMD Embedded (Xilinx): profitable through the trough, modest growth returning
— Intel Altera FPGA — now 51% private equity, no longer in the income statement
— The Vera CPU — a teaser worth watching as 2026 develops

Sponsored by fiscal.ai — 25% off any paid plan through May 14 only. Use our link: fiscal.ai/csi

Disclosure: Nick and Kasey hold positions in AMD. They do not currently hold Intel. This content is for general information only and is not individual investment advice. All investing involves risk.

chipstockinvestor.com

Almost nothing in semiconductor land is cheap right now. First Solar might be the exception — and that is worth paying attention to, even if you have never thought of a solar panel manufacturer as a chip stock.

First Solar trades at 12.7x forward earnings and 12.7x forward free cash flow. It carries over $2.4 billion in cash with almost no debt. It just reported record Q1 2026 revenue of just over $1 billion, up 24% year over year, with expanding profit margins. It manufactures domestically in five US facilities — a sixth is under construction in South Carolina. It benefits from Inflation Reduction Act tax credits through 2029. And it has just launched a new manufacturing process called CuRe — copper replacement — that increases panel lifetime energy yield by up to 8% and extends panel lifespan.

Everything checks out. Until you look at the guidance and the backlog.

Full year 2026 revenue is expected to come in flat to slightly down versus 2025. The order backlog, which peaked above 70 gigawatts in 2023, has now declined to under 48 gigawatts. First Solar is working through existing orders faster than it is winning new ones. A sizable cancellation from customer LightSource BP in 2024 and 2025 accelerated that decline. These are the hallmarks of a value trap — a stock that looks cheap because the future earning power is genuinely uncertain, not because the market has mispriced it.

The potential inflection point is a pending Section 232 investigation into whether crystalline silicon solar panel imports — primarily from China, where state-subsidized price dumping has been a recurring competitive tactic — constitute a national security risk. If the ruling lands in Q2 2026 and tariff protections follow, First Solar's order book could refill rapidly. If it does not, the backlog decline continues and the cheap valuation has every reason to stay cheap.

CSI walks through the full picture: the thin-film cadmium telluride technology edge, the CuRe manufacturing upgrade, the domestic supply chain advantage, the backlog reality, and a reverse DCF that shows the bar First Solar needs to clear is genuinely low — only 7% annual profit growth over five years with a 0% terminal rate gets you to today's price. The conclusion is honest: mildly interested, but the hallmarks of a value trap are present. Patience is the strategy.

What we cover:
— Why a solar company qualifies as a chip stock — and where First Solar fits in the supply chain
— First Solar Q1 2026: $1B+ revenue, record margins, $2.4B cash, minimal debt
— Thin-film cadmium telluride vs. crystalline silicon — the technology difference that matters
— The CURE manufacturing process: launching now in Ohio, targeted across all facilities
— Domestic US manufacturing as a competitive and geopolitical advantage
— The guidance problem: flat to down revenue in 2026
— The backlog decline: from 70GW in 2023 to under 48GW and still falling
— Section 232 tariff investigation — the binary catalyst expected Q2 2026
— Reverse DCF: 12.7x earnings, 7% growth, 0% terminal rate
— CSI verdict: mildly interested, but patient — the value trap signs are real

Disclosure: Nick and Kasey do not currently hold First Solar. This content is for general information only and is not individual investment advice. All investing involves risk.

chipstockinvestor.com