AIToday
AI Stocks & MarketsTop Companies' AI MovesAI Business & IndustryTop Companies AIPublished: Sep 12, 2026, 06:30 JST2 min read

NVIDIA's Huang: AI buildout could lift chipmakers

NVIDIA's Huang: AI buildout could lift chipmakers

3 Key Points

  1. What happened

    NVIDIA CEO Jensen Huang said the AI factory buildout, "the largest infrastructure expansion in human history," is accelerating, and top-five hyperscaler capex is tracking to $1.3 trillion in 2027, with NVDA at $218.36.

  2. Why it matters

    Each chipmaker now has multi-year contracted revenue behind it: NVIDIA has $279B in Rubin memory supply commitments, Micron $100 billion from 16 agreements, SanDisk $93.9 billion from eight New Business Model deals.

  3. What to watch

    Whether the AI capex cycle keeps compounding through 2027 decides the $300, $1,750, and $2,500 targets. Watch NVIDIA's Q4 gross margin, which memory costs pressure to 71% to 72%.

WHO IT HITSInvestors holding or weighing NVDA, MU, and SNDK face a decision that depends less on near-term earnings and more on whether hyperscaler capex keeps compounding through 2027. Memory pricing trends matter to both the buyers and sellers of chips here.

Ask the AI about this article →

Summaries like this, in your inbox every morning.

Context & Analysis

The shift the article describes is that compute and memory now sit on the same critical path, rather than memory following demand at a distance. Jensen Huang's comment that the AI factory buildout is "the largest infrastructure expansion in human history" frames the scale, and the numbers behind that frame are concrete: top-five hyperscaler capex is tracking to $1.3 trillion in 2027, and Huang describes Vera Rubin's revenue per gigawatt rising to $40 billion versus $25 billion for Blackwell.

The three companies each tell a slightly different version of the same story. NVIDIA's case rests on the Vera Rubin ramp and $279B in supply commitments tied to Rubin memory procurement. Micron's rests on HBM4 and 16 Strategic Customer Agreements with roughly $100 billion in minimum contractual revenue, with CEO Sanjay Mehrotra saying DRAM and NAND demand significantly exceed supply and tightness will last beyond calendar 2027. SanDisk's rests on NAND turning into what the article calls a contracted cash machine, with eight New Business Model agreements carrying a minimum $93.9 billion in expected revenue.

The tension worth watching is that memory inflation pressures NVIDIA's gross margin to 71% to 72% in Q4 even as it lifts Micron and SanDisk. So the same force cuts both ways across this group, and how long that price pressure persists, and whether capex keeps compounding through 2027, is what decides whether the round-number targets hold.

FAQ
What did Jensen Huang actually say on the call?
"The buildout of AI factories, the largest infrastructure expansion in human history, is accelerating at extraordinary speed."
How much contracted revenue do these chipmakers have?
NVIDIA has $279B in supply commitments for Rubin memory procurement, Micron has roughly $100 billion from 16 Strategic Customer Agreements, and SanDisk has $93.9 billion from eight New Business Model agreements.
What would it take to hit the round-number targets?
It requires the AI capex cycle to keep compounding through 2027, but each name already has contracted revenue, expanding margins, and rising estimates behind it.
Top Companies AIRead Original Article

Get the latest AI Stocks & Markets news every morning

For example, today's edition would include:

  • Intel-backed Kepler Computing claims it can beat HBMYahoo Finance AI · 1h ago
  • d-Matrix's Raptor to ride NVLink Fusion into Nvidia racks by 2027Yahoo Finance AI · 4h ago
  • JPMorgan cut off Situational Awareness lending after AI losses, source saysTop Companies AI · 7h ago

AI-summarized, only the topics you pick — one digest a day via Email, Slack, or Discord.

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. Q&As are published on this page for other readers too.

Related Articles

Next articleMeta's Muse AI pulls Amazon data, unnerves tester