In September 2025 Nvidia agreed to buy $5 billion of Intel common stock and to co-develop several generations of data centre and PC products with it. Intel will build custom x86 CPUs for Nvidia's AI platforms and x86 PC chips that integrate Nvidia RTX GPU chiplets, linked over Nvidia's NVLink. The deal gives Intel capital and a marquee partner, gives Nvidia a way into the x86 world it does not control, and puts pressure on AMD and on Arm-based PC chips.
This analysis was written in September 2025, shortly after the announcement, with a dated update on the deal's completion.
Key takeaways
- The deal: Nvidia buys $5 billion of Intel stock. The companies will co-develop AI server platforms and AI PCs that pair Intel CPUs with Nvidia GPUs over NVLink.
- Why it matters: it could ease supply bottlenecks, give "AI PC" real meaning, and reduce Nvidia's dependence on a single manufacturing partner.
- Who should worry: AMD, squeezed on both CPUs and GPUs; TSMC, if Intel takes on packaging or manufacturing work; Arm-based PC chips from Qualcomm and others.
- Risks: Intel's record of delays, geopolitics, and the chance that consumers shrug at AI PCs.
When rivals need each other
In technology, rivalry usually drives progress. Sometimes, though, survival or opportunity calls for working together.
In 1997 Apple was close to collapse and accepted a $150 million investment from Microsoft, its fiercest competitor. The crowd at Macworld booed. Steve Jobs told them Apple had to let go of the idea that for Apple to win, Microsoft had to lose. The deal steadied Apple and set up its recovery.
In 2011 Samsung and Apple were fighting hard in court over smartphone patents, yet Apple still relied on Samsung to make the processors inside the iPhone. That was necessity rather than hypocrisy.
In September 2025 another unlikely pairing arrived. Nvidia, the world's most valuable chip company, agreed to invest $5 billion in Intel, the former leader now working to win back its position. The two will develop several generations of products for data centres and PCs together: Intel will build custom x86 CPUs for Nvidia's platforms and x86 PC chips that include Nvidia RTX graphics, connected over NVLink.
This is more than a supply deal. It is a bet on how the semiconductor ecosystem, global supply chains and AI computing will develop.
The deal terms
| Item | What was announced or reported |
|---|---|
| Announced | 18 September 2025 |
| Investment | $5 billion of Intel common stock at $23.28 per share |
| Shares | About 214.7 million, as reported from regulatory filings |
| Data centre | Intel builds Nvidia-custom x86 CPUs that Nvidia integrates into its AI infrastructure platforms |
| PCs | Intel builds x86 system-on-chips that integrate Nvidia RTX GPU chiplets |
| Interconnect | Nvidia NVLink between the CPU and GPU parts |
| Not included | No commitment for Intel Foundry to manufacture Nvidia GPUs was announced |
Update (dated): the US Federal Trade Commission cleared the investment in early December 2025, and a filing on 29 December 2025 showed that Nvidia had completed the purchase, as reported by Tom's Hardware and other outlets at the time.
The deal also came weeks after two other investments in Intel in August 2025: SoftBank agreed to buy $2 billion of Intel stock, and the US government converted CHIPS Act grants into an equity stake of roughly 10 percent. Together they made Intel's balance sheet a national industrial policy story as much as a corporate one.
What each side brings
If this were only a rescue, it would matter less. But Nvidia and Intel bring strengths that fit together and could change the PC and data centre platform market.
Nvidia
- AI compute leadership. Its GPUs are the default choice for training and inference. "Do you have H100s?" became the first question in AI infrastructure.
- The CUDA software base. CUDA and its libraries, such as cuDNN and TensorRT, are deeply embedded with developers. Once you build on CUDA, switching is expensive.
- NVLink and system design. Nvidia sells complete systems such as DGX and HGX with its own interconnects, not just chips.
- Brand. For companies and consumers alike, Nvidia now means AI acceleration.
- Financial strength. With gross margins typically in the 70 percent range in its recent fiscal years, as reported in its results, Nvidia can fund strategic bets like this easily.
Intel
- CPUs everywhere. Despite AMD's gains, Intel's x86 CPUs still power most PCs and many servers.
- Advanced packaging. EMIB, Foveros and years of investment make Intel one of the few companies able to build chiplet designs at scale.
- US manufacturing. Intel is the only US company trying to run leading-edge fabs at scale, which makes it central to industrial policy.
- OEM relationships. Long ties with Dell, HP, Lenovo and every major PC maker. Intel knows how to launch platforms through OEMs.
- Control of the client platform. Because Intel supplies the CPU, it has a big say in which accelerators reach consumer PCs.
Where the partnership adds value
Together, the two can do things neither could alone:
- CPU and GPU working closely. Intel CPUs linked to Nvidia GPUs over NVLink can cut latency and raise throughput, which matters for cloud AI workloads.
- AI PCs that deliver. OEMs could ship laptops with RTX-class graphics built in, not just modest NPUs, which would give the "AI PC" label substance.
- More capacity. Nvidia needs more supply, and Intel's fabs and packaging could offer a second route.
- A common platform. Nvidia has the developers; Intel has the OEMs. Together they can bring both sides along.
- A geopolitical hedge. More US and European capacity, and less reliance on Taiwan, helps both companies with regulators and national security concerns.
How it looks across the ecosystem
1. PC and server makers: something worth selling
For years PC makers have offered small steps: thinner laptops, a little more battery life, slightly better NPUs. Buyers have had few reasons to upgrade.
With Intel and Nvidia together:
- AI PCs could run assistants and language models locally and speed up video, 3D and coding work in ways people notice.
- Servers could come in versions with a different CPU and GPU balance, so OEMs can offer systems tuned for inference or for training.
That could start the first real PC upgrade wave in years.
2. Cloud providers: more negotiating power
Microsoft, AWS, Google and Meta are Nvidia's largest customers and often its loudest critics. GPU shortages have left them paying more and waiting longer.
The partnership could give them:
- more capacity, if Intel packaging comes into play;
- custom designs tuned for their workloads;
- leverage, by setting Nvidia and Intel against AMD, TSMC-based or Arm-based alternatives.
For the cloud providers, this could be the first loosening of Nvidia's hold.
3. Developers and start-ups: more predictable supply
For AI start-ups, GPU shortages have been a real threat. Projects stalled for lack of hardware, not money.
With Intel involved:
- supply may become more predictable;
- consumer devices could get tighter CPU and GPU integration, widening the market for AI software;
- the same CUDA stack could run in the cloud and on AI PCs, which makes it easier to build products that span both.
4. Governments: chip sovereignty
This is a political story as much as a business one.
- United States: the CHIPS Act put billions into Intel's fabs, and Nvidia's investment supports that bet.
- Europe: Intel's fab in Ireland could play a part, in line with EU goals for local supply.
- China: will see this as more Western alignment that limits its access to leading AI hardware, and is likely to push harder on its own GPUs and interconnects.
Chips have become a strategic resource, and this deal shifts the balance.
5. Investors: a new story for Intel
Intel's share price has suffered from years of missed deadlines. Nvidia's stake changes the story:
- Intel gains outside validation;
- Nvidia looks pragmatic rather than complacent;
- AMD investors face pressure on both CPUs and GPUs.
6. Competitors: harder choices
- AMD: the most exposed. Its EPYC CPUs and Instinct GPUs face a combined Nvidia and Intel platform.
- TSMC: still essential, but Nvidia exploring Intel weakens its position.
- Arm and Qualcomm: their AI PC plans look less certain if OEMs back Intel and Nvidia.
- Broadcom and network vendors: NVLink-first designs may sideline open standards such as CXL and deepen Nvidia lock-in.
7. Consumers: the quiet winners
Consumers do not follow foundry yields, but they notice when a laptop handles AI tasks without the cloud. If the partnership delivers assistants that work offline, much faster creative tools and gaming and AI in one machine, ordinary buyers will feel it, and a market many saw as stagnant could grow again.
Who benefits
- OEMs: a way to stand out beyond commodity hardware.
- Cloud providers: more bargaining power on price.
- Start-ups: steadier GPU supply.
- Governments: support for domestic chipmaking.
- Investors: a new growth story.
- Consumers: useful AI features rather than gimmicks.
The bear case
- Intel misses its targets again.
- Consumers see AI PCs as a gimmick.
- Geopolitical tension disrupts the partnership.
- Nvidia keeps its most important chips at TSMC, and Intel's role stays small.
The bull case
- Intel scales up packaging and eases bottlenecks.
- AI PCs start the first global PC upgrade cycle in a decade.
- US and EU policy support the alliance with funding and contracts.
- Cloud providers adopt the new systems and diversify their infrastructure.
Historical parallels
- Apple and Microsoft (1997): survival through investment.
- Apple and IBM (2014): enterprise credibility through partnership.
- Samsung and Apple (2010s): rivals in stores, partners in supply.
The pattern: big platform shifts create unlikely alliances.
What to watch next
- OEM launches: will Lenovo, Dell and HP commit to AI PCs with Nvidia graphics?
- Packaging results: can Intel show volume and yield?
- Manufacturing: any sign of Nvidia GPUs built in Intel fabs? This was not part of the September 2025 announcement.
- User experience: do AI PCs genuinely help people, or just tick marketing boxes?
Closing thought
The Nvidia and Intel partnership is strategy rather than charity. Nvidia secures supply and strengthens its platform. Intel gains a major partner and a better story.
The effects reach much further: OEMs, cloud providers, start-ups, governments, investors and consumers. This is an ecosystem move as much as a deal between two companies.
The most important alliances in tech have come from necessity, not friendship. Microsoft once helped save Apple. Samsung made chips for the iPhone it competed against. Now Nvidia may have helped save Intel, not out of nostalgia, but because the future of AI computing needs it.
FAQ
How much of Intel does Nvidia own?
About 214.7 million shares. Reports put this at roughly 4 to 5 percent of Intel; the exact share depends on Intel's share count at the time, so treat the percentage as approximate.
Will Intel make Nvidia's GPUs?
Not under the announced terms. The agreement covers CPUs and PC chips from Intel combined with Nvidia GPU technology. Foundry work for Nvidia was not announced.
What is NVLink and why does it matter here?
NVLink is Nvidia's high-bandwidth, low-latency link between processors. Using it between an Intel CPU and an Nvidia GPU, instead of only PCIe, lets the two share data faster, which helps AI workloads that move large tensors between host and accelerator.
What does this mean for AMD?
AMD is the only company selling both x86 CPUs and data centre GPUs at scale. A joint Intel and Nvidia platform competes directly with that combined offer.
Sources
- NVIDIA and Intel to Develop AI Infrastructure and Personal Computing Products, joint press release, 18 September 2025
- Tom's Hardware, Nvidia gives Intel a lifeline with $5 billion common stock deal (FTC approval), December 2025
- Intel and SoftBank, and US Department of Commerce announcements on Intel equity investments, August 2025
Your comments will be moderated before it appears here.