Nvidia Promises to Sell Twice as Many Chips in 2027 After a $96 Billion Quarter

Jensen Huang rarely gives numbers for the following year. On September 17, 2026, at an event in the United Kingdom, the Nvidia CEO did: the company expects to sell twice as many chips in 2027 as in 2026, driven by AI investment “across industries and countries.” The remark lifted the stock 2% the next day and comes on top of a quarter that was already a record — $96.2 billion in revenue, up 106% year over year. Here are the numbers behind the promise and what could get in the way.

Quick answer: what did Huang say?

That the physical volume of chips Nvidia sells should double in 2027. In August, the company had already projected revenue growth of 70% for the next fiscal year and said it could double if supply were sufficient. The new statement is about units, not dollars: it suggests hardware expansion may be even faster than revenue growth. For the current quarter, guidance is roughly $108 billion.

The numbers behind the promise

Metric	Value	Change
Fiscal Q2 revenue	$96.2 billion	+106% year over year
Data center revenue	$89 billion	+117%
Q3 guidance	~$108 billion	—
Revenue projection (next fiscal year)	+70% (or double, with enough supply)	August 2026
Chip volume in 2027	2x versus 2026	Sept 17 statement

Nearly all of the growth comes from data centers: $89 billion of the $96.2 billion. That is money from OpenAI, Anthropic, Google, Microsoft, Meta and the governments building “AI factories.”

Where the confidence comes from

Vera Rubin and the per-megawatt metric

Nvidia’s next platform, Vera Rubin, showed up to seven times more tokens per megawatt than Blackwell in tests. With power now the physical limit of data centers, a chip that does more with the same electricity sells more units without waiting for new power plants.

Contracted demand

Infrastructure startups such as Crusoe ($3.9 billion raised this week) and Figure AI (up to 100,000 GPUs contracted) show the buyer queue extends beyond big tech. And Broadcom, its rival in custom chips, grew 86% over the same period — the whole pie is growing.

What could go wrong

Three risks show up in Huang’s own words. Supply: doubling units depends on TSMC, HBM memory and power — the company has already admitted revenue “could double if supply were sufficient.” Wall Street: the same week, analysts questioned the return on AI infrastructure spending, and sector stocks wobbled. Geopolitics: Huang attended the state dinner with Xi Jinping, but export restrictions to China remain, and Huawei just pulled its next AI chip forward by nine months.

Why this matters to you

For investors, the promise to double units is the most aggressive figure Nvidia has ever given publicly — and it will be scrutinized quarter by quarter. For businesses that buy cloud, more chips mean more capacity and, over time, a lower price per AI token. And for anyone following the industry, the “bubble or not” question of 2026 will be settled by a simple number: if Nvidia ships twice the chips and customers keep buying, it was not a bubble.

Frequently asked questions

How much did Nvidia earn last quarter?

$96.2 billion in fiscal Q2, up 106% year over year, with $89 billion from data centers (+117%). Guidance for the following quarter is about $108 billion.

What did Jensen Huang promise for 2027?

That Nvidia will sell twice as many chips in 2027 as in 2026, a statement made on September 17, 2026 at an event in the UK.

Why did Nvidia stock rise?

Shares gained about 2% after Huang’s remark about doubling chip volume, a demand signal that goes beyond the revenue projections released in August.

At DigitalRadar, we follow the giants that move technology. Stay on the radar so you do not miss the next analysis.