Every AI boom rests on layers. One layer builds the tools. Another builds the models. A third turns those models into products customers actually pay for. Wall Street spent the last two years obsessed entirely with the first layer – chips, cloud infrastructure, data centers, power generation – and the trade worked spectacularly. Nvidia, AMD, Oracle, and a wave of infrastructure names became the market’s biggest winners simply by supplying the buildout.
Table of Contents
Anthropic sits in the second layer. That’s why this IPO matters more than its size suggests.
The company hit a $61.5 billion valuation in its latest round, with annualized revenue reportedly approaching $3 billion and profitability finally entering the conversation. Anthropic itself isn’t the story here, it’s the catalyst.
The real question is whether companies building large language models can generate revenue commensurate with the hundreds of billions flowing into the infrastructure built to serve them. Investors spent two years buying the suppliers without ever testing the demand side. Anthropic’s IPO hands them the first real chance to do exactly that, and Oracle and AMD sit directly in the blast radius of whatever answer the market delivers.
Oracle Bet $138 Billion On Demand That Hasn’t Arrived Yet
Last quarter, Oracle’s Remaining Performance Obligations hit $138 billion against fiscal 2026 revenue of $57.4 billion. That’s customer commitments running more than double a full year of revenue, with enterprises reserving cloud capacity years before they’ll actually need it. Cloud Infrastructure revenue climbed 52% year-over-year. OCI consumption grew 62%. Management is guiding OCI growth above 70% for fiscal 2027, and guidance at that level only makes sense if the people setting it believe the demand curve keeps climbing rather than flattening out.
Now, Oracle Corp (NYSE: ORCL) sits one layer beneath Anthropic in this stack. The models generate the headlines; Oracle’s infrastructure processes the workloads underneath them. Every enterprise rolling out Claude, every workflow rebuilt around it, every increase in model usage eventually becomes compute, storage, and networking demand flowing straight back to Oracle. Or put another way, today’s infrastructure spend assumes tomorrow’s workloads show up on schedule.
Every customer Anthropic adds and every dollar of revenue it generates strengthens the case that Oracle’s backlog reflects real demand instead of excessive optimism. The relationship runs both directions, if model builders stall, the infrastructure built underneath them gets questioned next.
The chart shows the market already wrestling with that tension. Shares ran toward $250 before falling back to roughly $165, now sitting below the 20-day moving average of $203, the 50-day near $190, and the 200-day around $204. The cloud numbers are among the strongest Oracle has ever posted. What’s being priced instead is whether that infrastructure finds enough demand to fill it, and Anthropic’s IPO delivers one of the clearest signals the market will get this year.
AMD Sells The Hardware That Makes The Bet Real
Oracle sells capacity. Advanced Micro Devices Inc (NASD: AMD) sells the silicon that capacity runs on, and every AI company eventually collides with the same wall regardless of how good its models are: compute.
AMD posted first-quarter revenue of $10.3 billion, up 38% year-over-year. Data Center revenue hit $5.8 billion, growing 57%. Segment operating income climbed 72% to $1.6 billion, and free cash flow reached a record $2.6 billion… numbers describing an industry consuming compute at a pace that barely existed eighteen months ago.
Management provided another figure that helps explain the scale.
Meta plans to deploy up to six gigawatts of AMD Instinct GPU capacity across future builds. Utilities use gigawatts to talk about powering cities and industrial corridors. AI infrastructure now speaks the same language, and that alone tells you how far past “tech trend” this buildout has already moved.
Anthropic operates inside that exact supply chain: Oracle supplies the infrastructure, AMD supplies the hardware, Anthropic consumes both, and every increase in model usage creates demand for accelerators, networking gear, and the cooling and power systems wrapped around them. AMD’s stock reflects institutional conviction that this chain holds together: shares trade near $520, above the 20-day moving average around $510, the 50-day near $423, and the 200-day around $265. That spread between current price and longer-term trend is a market betting heavily that AI demand keeps expanding. Anthropic’s IPO either validates that bet or forces a serious repricing of it.
This IPO Tests The Entire Ecosystem, Not One Company
Most IPOs like Saudi Aramco, VISA, Amazon etc, reveal something about each company going public. This one reveals something about everyone standing behind it. Oracle’s backlog assumes AI demand keeps compounding. AMD’s valuation assumes model builders keep spending aggressively. Data-center operators keep expanding capacity, utilities keep building new power infrastructure, and capital keeps flowing into every corner of this ecosystem because investors believe large language models eventually generate enough economic value to justify what’s being spent around them.
Anthropic may become the first major public report card on that belief. A strong reception confirms model builders are converting growth into durable businesses and keeps capital flowing into cloud infrastructure and chip production exactly as planned. A weaker one shifts the conversation toward margins, returns on invested capital, and a credible profitability timeline… demands that adoption metrics alone won’t satisfy anymore. In all of these, infrastructure providers keep their existing contracts regardless of which outcome plays out. What changes is how much confidence the market keeps extending to the assumption that demand grows forever.
AI creating real value is not the debate here. What we need to see now is revenue showing up, profitability following it, and returns on capital justifying the money that built this entire stack.
Leave a Reply