Nvidia connects the puzzle pieces: why one quarterly report should set the tone for the market

Reading earnings reports is like solving a puzzle: the edges provide some guidance, but the picture only becomes clear once the key pieces fall into place. Nvidia is such a key piece. With its quarterly results and outlook, the company provides not only insight into its own performance but also into the broader dynamics of the technology sector. Nvidia sets the pace: where it goes, the market follows.

That central role is due not only to Nvidia’s status as the world’s largest listed company (over $4.4 trillion in market capitalization, larger than Apple and Microsoft), but also to its unique position in the AI value chain. Where hyperscalers invest billions, those amounts flow directly into Nvidia’s revenues.

With the exception of Tesla, the other Big Tech players reported solid results in recent weeks. Microsoft’s Azure cloud revenue accelerated to 39% growth, Alphabet reported a rising backlog of cloud contracts and a recovering advertising market, while Meta boosted ad revenues by double digits. Signals that should, in principle, encourage investors.

Yet the market has reasons to hesitate. The very companies posting strong results have been ramping up their AI investments at an unprecedented pace since late 2023. If Microsoft maintains its current quarterly spending rate of $30 billion, that would amount to $120 billion annually. Alphabet plans to invest $85 billion, Meta about $70 billion, and Amazon $120 billion. The bulk of this goes to data centers and AI infrastructure, though Microsoft stressed that more than half is being invested in long-lived assets such as buildings and energy systems. Only part flows directly into servers, CPUs and GPUs.

The optimistic interpretation: this investment wave reflects the immense potential that AI could unlock in the years ahead. The pessimistic: Big Tech is caught in an AI arms race that could end in overcapacity, echoing previous episodes of technological overinvestment.

AI Is not Dotcom 2.0

Today’s AI boom reminds some of the dotcom hype of 2000. Back then, telecom giants like AT&T, Verizon and Deutsche Telekom poured billions into fiber optics and dark fiber, far beyond real demand. Equipment makers such as Cisco and Lucent saw record sales, but the investment wave ultimately proved unsustainable.

That wave did, however, lay the foundation for the rise of internet companies. Some, like Amazon, eBay and Google, became lasting successes. Others, like Pets.com and Boo.com, collapsed spectacularly. Cisco’s valuation soared, but it was the implosion of countless start-ups and loss-making listed companies that scarred investors and undermined confidence in the tech sector.

For pessimists, today’s AI market looks like a repeat: hyperscalers are pouring hundreds of billions into AI infrastructure (data centers, GPUs, memory chips and networks), and Nvidia plays the role Cisco once did. The difference: Nvidia is in a much stronger position. Around this hardware layer, a thriving AI ecosystem is emerging. Foundation model builders such as OpenAI and Anthropic, newcomers like Mistral and Cohere, and established software firms like Salesforce, Adobe and ServiceNow are all pushing into AI. Platforms such as MongoDB, Palantir, Datadog, Elastic and Snowflake are gaining ground, while pure AI companies like C3.ai, Scale AI and Hugging Face are building their place in the value chain. And this is only the beginning: hundreds of AI start-ups and scale-ups are ready to break through.

Dotcom vs. AI: the numbers

In words, the comparison with 2000 may sound plausible, but the numbers show a fundamental difference. The only thing telecom operators then shared with today’s hyperscalers is a combined revenue base of about $1.2–1.4 trillion. Back then, that amount was split among some 30 large players (80% of investments) and over 200 smaller operators (20%).  Profitability was limited, and most carried heavy debt.

Dotcom period (2000)

By contrast, today’s AI infrastructure companies generate far higher revenues and, crucially, margins two to three times greater. They also enjoy robust cash positions and strong cash flow generation.

AI period (2025)

The biggest difference lies in the third category. In 2000, most internet companies were start-ups without viable business models. Today, many are established, profitable firms integrating AI into their operations. New AI start-ups also tend to launch with clearer business models and real revenue from day one. As then, venture capital is abundant, but it will be important to avoid excesses in the years ahead.

AI spending is exploding, but the market may be underestimating its impact

Since 2022, operating cash flows at Microsoft, Meta, Alphabet and Amazon have been rising steadily, while capital expenditures in data centers, AI infrastructure and networks have more than doubled since 2023. Despite this surge in investments, free cash flows remains clearly positive. That said, growth has stagnated since 2023, with a slight decline expected this year.

 

 

Nvidia is the missing puzzle piece to complete this picture. Once its numbers are added, the scale and dynamics of this investment wave become clearer. Three things stand out:

1. Operating cash flow growth accelerates further, boosted by Nvidia.
2. Free cash flow rises in 2024 and continues to grow slightly in 2025.
3. From 2026, a turning point: free cash flows outpace capex and move above it.

 

 

What the chart does not show, but is crucial: the gap between investment expectations and revenue forecasts. consensus still expects a doubling of AI investments the MAG 5 while projecting slowing revenue growth, from high-teens now to low-teens in coming years.

• Skeptics see overcapacity risk.
• Optimists see forecasts that are too conservative: if AI adoption scales, revenues could beat expectations.

Nvidia’s results suggest the latter: this wave is not hype but a structural reshaping of value chains, a fundamental shift laying the foundation for the next phase of digital growth.

Nvidia’s results in detail

The figures Nvidia published on Wednesday once again underline why the company plays such a pivotal role in the current investment wave. In the second quarter, revenue reached $46.7 billion, up 57% year-on-year and 6% quarter-on-quarter. That is more than $17 billion above the $30 billion of the same quarter last year. To put this in perspective: that additional revenue in just one quarter equals more than twelve times the annual turnover of a company like Lotus Bakeries.

Quarterly revenue came in above both Nvidia’s own guidance and the official market consensus. That consensus is compiled from the formal estimates of 67 sell-side analysts covering the stock. On Wall Street, however, there is also always an informal “whisper number,” a kind of bidding game in which expectations are pushed ever higher. Against that backdrop, even strong quarterly results can sometimes feel less spectacular. This explains the rather muted share price reaction. As I summarized on Kanaal Z on Thursday: there was a beautiful red cherry on the cake, but the fireworks were missing. Better than expected, but not as high as hoped.

The core driver remains the datacenter division, which generated $44.1 billion, up 56% from a year ago. The large cloud players, Microsoft, Alphabet, Amazon and Meta, together still account for about half of revenues, though their relative share is gradually declining. New customer groups are emerging:

• Governments such as Saudi Arabia, Abu Dhabi, Japan, Norway and France, which are expected to contribute around $20 billion, or about 10% of revenue this year.
• Neo-cloud players such as CoreWeave, Nebius, Stargate, Lambda Labs and Crusoe Cloud, now contributing an estimated 15%.
• Enterprise and consumer-facing companies like Disney, Foxconn, Hitachi and Bytedance, which are also embracing Nvidia’s technology more broadly.

Profitability is equally impressive. The gross margin rose to 72.4%, far above the sector average. Operating profit reached $30.2 billion, up more than 51% year-on-year. Net income came in at $26.8 billion, more than double the same period last year. Free cash flow for the quarter was $15.4 billion, higher than a year earlier thanks to strong revenue growth, though partly offset by working capital effects related to cash collection timing and the ramp-up of the Blackwell chip. For the full year, Nvidia is on track to generate $100 billion in free cash flow. That explains why the board decided on August 26 to allocate another $60 billion for share buybacks.

In short: excellent results, but the missing fireworks had everything to do with China. Revenues from China declined year-on-year; without that drop, growth would have been around 67% instead of 56%. China also weighed heavily on the outlook. Nvidia still projects an exceptional growth path, guiding for third-quarter revenue of $53–55 billion, a year-on-year increase of about 57% or +$20 billion. This was above market expectations, but it could have been more.

The reason lies in the tense U.S.–China trade relationship. In April, Nvidia was barred from shipping its most advanced AI chips to China. Initially, this affected the Hopper generation; now the new Blackwell chips are also restricted. Nvidia developed a lower-performance variant for the Chinese market (the H20), but in June that chip too was hit by an export ban. Then, in July, Washington shifted course again, starting a review of licenses for H20 sales to Chinese customers. A small number of Chinese buyers have since been granted licenses, but Nvidia says no shipments have yet taken place.

Remarkably, the U.S. government has floated the idea of taking a 15% cut of licensed H20 sales, though this has not yet been formalized in regulation, a measure with a clear Trump-like scent. Nvidia therefore excluded H20 sales from its third-quarter guidance. Should a breakthrough in the geopolitical standoff occur, the company says it could immediately recognize $2–5 billion in additional quarterly revenue from already signed contracts. Nvidia’s CFO added pointedly: “if we had more orders, we could bill more.” This seems stating the obvious, but basically means the $2–5 billion could have been more. The remark implicitly refers to Beijing’s counter-move: reportedly pressuring its domestic AI companies not to buy Nvidia chips after sharp comments from U.S. Commerce Secretary Howard Lutnick. CEO Jensen Huang sketched the bigger picture: the Chinese market alone represents $50 billion in potential this year, with the prospect of ~50% annual growth. Still, no reason for despair. With or without China, Huang hit the nail on the head:

“We are at the beginning of an industrial revolution that will transform every industry. We see $3 to $4 trillion in AI infrastructure spend by the end of the decade. The scale and scope of these build-outs present significant long-term growth opportunities for NVIDIA.”

That context explains why Nvidia’s valuation remains elevated. At around 40 times earnings, the stock looks expensive compared with the broader market. But once strong growth prospects and high visibility are factored in, the premium looks far more defensible. The numbers show clearly that Nvidia not only rides the wave of hyperscaler investments, it is actively steering them.

 

Bron: Online Press Kits – NVIDIA Newsroom (2025)