Intel: Embracing Nvidia's 'big thigh', who is the real winner?

As the undisputed major event of the year in the American chip industry, after the U.S. government exchanged equity for subsidies of over $8 billion and SoftBank made a symbolic investment, last night finally secured a $5 billion investment from NVIDIA. The American version of 'SMIC' has finally found a 'big leg' sponsor.

So in this transaction, who wins and who loses, and who will have the last laugh? Let's take a detailed look at the ins and outs of this matter:

I. Sudden Event: NVIDIA's $5 Billion Investment in Intel

On September 18, 2025,$NVIDIA(NVDA.US) and$Intel(INTC.US) announced a collaboration to jointly develop multiple generations of customized data center and PC products to accelerate applications and workloads in hyperscale, enterprise, and consumer markets.

Details are as follows:

① In the data center field, Intel will customize x86 data center-specific CPUs with NVlink interfaces for NVIDIA, which NVIDIA will integrate into its AI infrastructure platform and market;

② In the PC field, Intel will launch x86 system-on-chip (SoC) integrated with NVIDIA RTX GPU chips.

③ NVIDIA will invest $5 billion in Intel's common stock at $23.28 per share (a 6.5% discount compared to the stock price the day before the transaction announcement).

II. Background: The 'Unhelpable' Intel

Intel, which was once glorious in the PC era, has fallen to the point of needing repeated joint investments from giants? In Dolphin Research's view, the core issue is that in the cloud and AI era, Intel has been substantially 'pierced by three arrows':

① 'Kicked out' of the data center battlefield:

Data center GPU: Intel completely missed the market opportunity for GPUs in the AI wave, having no presence in the data center market;

Data center CPU: Due to increased GPU investment, CPU updates are slow, and Intel's core customers—cloud vendors—are themselves extending the depreciation cycle of traditional CPU servers, weakening the demand for new CPU purchases.

Intel's own strong point—data center CPU servers—failed in GPU products, as competitors AMD and NVIDIA bundled their ARM architecture CPUs with their GPU sales, offering rack and system integration solutions, causing Intel's market share in the data center CPU market to fall from the original 80-90% to around 75%.

As a result, normal data center suppliers are expanding in the AI era, but Intel is completely stagnant in the face of cloud vendors' capital expenditure waves.

② PC Market:

Intel is also significantly impacted by AMD in the PC market, especially in the desktop CPU market, where AMD has surpassed Intel in market share.

Client business is the company's traditional advantage and main source of profit. If the company loses even this basic market, it will be difficult to bring hope to the market.

Meanwhile, in the upcoming AI PC prospects, because ARM architecture instructions are relatively simple, running GPU, CPU+NPU is faster and more energy-efficient, as ARM architecture computer performance continues to improve, Intel's position in the traditional computer CPU market will face greater threats once fully entering the AI PC era;

③ Slow progress in external foundry: Under the heavy asset, vertically integrated layout of chip self-research and self-production, Intel's internal CPU business is hopeless, orders are shrinking, and foundry is gradually falling behind.

In terms of external foundry progress, it has not gained market trust, and the company hopes to achieve breakthroughs at the 18A node. In terms of transistor density, 18A is still significantly lower than TSMC's N3P, and the market itself accepts its lag.

For Intel's 18A, the market mainly focuses on whether it can be used. Currently, 18A has not won market trust. The company only plans to adopt the 18A process for its Panther Lake first and observe the market's reaction.

The result of the resonance of the above (①——③) is that since 2022, Intel has been continuously facing operational losses. Although there is still $9.6 billion in cash and equivalents on the books, it cannot withstand continuous losses. If the loss rate is $4 billion/year, this money can only last for about 2 years.

In the absence of improvement in company business, the company can only start the rhythm of layoffs, cost reduction, and asset sales to reduce the extent of losses. The company's performance has been struggling at the bottom, stock prices continue to decline, and the company's PB level has fallen to around 1.

III. Background: The 'Endless Father'—U.S. Government

The problem is that after entering the cloud and AI era, although Intel's asset quality is not great, its foundry business is a scarce asset for the U.S. government, as in high-end 'logic chip manufacturing,' it is almost the only one. Other U.S. semiconductor manufacturing assets, whether Micron, Texas Instruments, etc., are more focused on slightly lower-end storage and automotive chip manufacturing.

Therefore, Intel has become the main target of U.S. government subsidies, receiving over $8 billion in subsidies (of which $5.7 billion has been received, and $3.2 billion is pending Trump administration approval), becoming a substantial 'subsidy king,' forming an 'unhelpable' entity, paired with an 'endless father' U.S. government.

However, the Trump administration believes that this endless 'blood pumping' is not a solution, so it adjusted its strategy, using subsidies to exchange for equity, thereby forming substantial influence over the company.

Since August, Intel has carried out a series of operations, introducing SoftBank, the U.S. government, and NVIDIA as company shareholders, and selling Mobileye shares.

The way the U.S. government obtained equity is very interesting. The money it invested was actually the $5.7 billion subsidy previously allocated under the Biden administration's 'CHIPS Act,' plus the pending $3.2 billion subsidy, totaling $8.9 billion, in exchange for a 9.9% equity stake.

After this round of financing operations, reorganizing the current equity structure of Intel, it can be seen: the original Vanguard Group and BlackRock are more in the form of financial investments, Intel has no controlling shareholder, but the single largest shareholder has become the U.S. government. In other words, Intel has become a U.S. state-owned enterprise.

Everyone should remember that in previous earnings report reviews, Dolphin Research compared Intel to the American version of 'SMIC,' and in fact, after Trump's subsidy-for-equity operation, the relationship between the enterprise and the government has become even more similar!

Previously, a large amount of investment in SMIC mainly came from 'blood transfusions' from the government and state-owned enterprises. In fact, both countries are continuing the demand for 'local production and independent control of the semiconductor industry.'

Even, the involvement of SoftBank and NVIDIA is also full of 'government father' orchestrated drama: the pricing of NVIDIA's investment reveals clues—from NVIDIA's subscription price of $23.28 to SoftBank's subscription price of $23, both are discounted based on the previous day's trading price.

Generally speaking, assets actively sought by investors are usually acquired at a premium, while only deals where the seller actively seeks money are usually accompanied by discounted sales of the target.

IV. Can Intel Really Turn Around and Head Straight for the Bright Road?

After recent capital operations, Intel not only changed its CEO but now also has deep-pocketed major sponsors like the U.S. government, SoftBank, and NVIDIA. After such operations, even if Intel continues to lose money in operations, it can afford it, directly dispelling the situation of Intel falling into 'bankruptcy valuation' again.

Compared to the previous U.S. government equity investment, NVIDIA's investment brings more confidence. NVIDIA not only provides financial support but also brings new opportunities to the business, with the chance to reverse the declining situation on 'three fronts.'

① Data Center Main Battlefield: In the current hot AI situation, Intel's quarterly revenue from data centers has remained at around $4 billion, indicating it has not entered the AI main battlefield. In contrast, NVIDIA is itself a leader in the data center market.

Previously, the interconnection between NVIDIA GPUs and x86 CPUs was mainly based on PCIe protocol. After the collaboration, the dedicated x86 CPU integrates NVIDIA's NVlink interface, directly increasing interconnection bandwidth speed by 14 times, and directly integrated into NVIDIA's Rubin and Vera architecture cabinets, effectively helping Intel's CPU to re-enter the AI data center main battlefield.

② CPU Market: Intel plans to launch x86 SoC (integrated system chip) with integrated NVIDIA RTX GPU graphics cards, which can be used for AI PCs and workstations, also bringing a competitive rebound for Intel and x86 systems in the traditional PC market, competing against AMD's integrated graphics chip.

③ External Foundry Customers: NVIDIA's foundry is mostly at TSMC, while Intel's external foundry progress has been slow. Of course, NVIDIA stated that before 2027, it may not launch new products from the collaboration, and the chips from the collaboration will not be manufactured at Intel's factories, but it does not rule out the possibility of future cooperation in foundry.

Previously affected by continuous 'low' business, Intel reduced its capital expenditure from last year's $24 billion to this year's $18 billion, which is roughly the level of maintenance capital expenditure. After becoming a U.S. 'state-owned enterprise,' Intel may also gain some confidence in 'expanding investment.'

Overall, this collaboration cannot guarantee that Intel will definitely revive, but at least through NVIDIA's assistance, it has gained a chance to turn the tables, and considering the government truly values the foundry assets, but this time NVIDIA has not made any commitments on foundry, if there is progress in foundry in the future, Intel will have the opportunity for further revaluation.

The government itself brings capital into the game, while also bringing in industry giants, indicating that Intel is indeed an entity that cannot be moved even if it is rotten, with certain bottom value, and the government's entry at $20/share has to some extent provided an implicit valuation bottom from the government's attitude.

V. NVIDIA: Not Big in Direct Impact, Wins in Imagination Space?

For NVIDIA, on one hand, it can fulfill its commitment to the U.S. government to invest in American semiconductors, even if the investment ultimately fails, the $5 billion investment size is just small change for a company with a market value of over $4 trillion.

From a purely business perspective, there is no particularly significant impact:

Because whether it's data center CPUs or AI PC CPUs, it mainly provides customers with a more comprehensive and better experience and solutions, and given the current business size of the company, CPU revenue is unlikely to constitute effective revenue enhancement.

Direct comparison: Intel's latest quarterly total data center revenue of $4 billion vs. NVIDIA's data center revenue of $41.1 billion in the same quarter, even if Intel gives all its data center revenue to NVIDIA, it's just a small amount, not considered large.

In Dolphin Research's view, the impact on NVIDIA is not on the surface, perhaps more left for everyone to imagine. For example, will Trump in future national trade negotiations require these countries (Middle East, Europe, etc.) to purchase more advanced American semiconductor equipment, or even in trade with China, will it open a bigger door for NVIDIA, rather than the currently sold H20, RTX PRO 6000D versions with severe cuts and weak competitiveness?

VI. Derivative Shockwave: AMD Most Miserable, Capex Stocks Benefit

① AMD: Clearly negative impact. After being snatched by Broadcom for the 'second place' position in AI chips, it was stepped on again by the collaboration between Intel and NVIDIA. The two companies' collaboration to provide data center x86 solutions will also jointly enter the AI PC market, posing a danger to AMD's PC market.

② Semiconductor Equipment (ASML, AMAT, Synopsys, etc.): Potential positive impact. Although the collaboration between Intel and NVIDIA mainly focuses on x86 architecture and AI PC integrated SoC, it may still extend to the foundry field in the future.

Especially from the government's perspective, the core of rescuing Intel is the scarcity of wafer manufacturing, ultimately hoping Intel can eventually repair its wafer foundry capabilities

And at present, Intel previously affected by 'three-line defeat,' significantly reduced this year's capital expenditure. If subsequent business improves or foundry breakthroughs, there is potential for increased capital expenditure, relatively beneficial to upstream equipment manufacturers, especially Synopsys, AMAT, ASML, etc.

These companies, more or less, have been affected by Intel's reduced capital expenditure, impacting their own performance, stock prices stagnating, and even declining since 2025. Intel's ability to restart capital expenditure imagination opens up, and in the case of these companies' stock prices not being high, it can also bring effective boosts.

③ TSMC: Currently, the impact is not significant. This collaboration mainly focuses on x86 structure and integrated SoC, without involving significant foundry business, thus having little impact on TSMC's current foundry business. As the collaboration progresses, if Intel obtains large foundry orders or achieves breakthroughs in foundry, it may impact TSMC.

VII. Summary: Some Happy, Some Sad

In summary, Intel and NVIDIA's collaboration, Intel will be the biggest beneficiary. After the U.S. government, NVIDIA, and SoftBank become company shareholders, Intel is unlikely to see 'bankruptcy valuation' again. The extent of the company's subsequent recovery mainly depends on the specific performance of the collaboration products and breakthroughs in foundry business.

As for NVIDIA, although it is the main participant, it will not have a significant boost to its own performance, more imagination for future broader market business development.

AMD should be the most obvious negative impact, as Intel and NVIDIA join forces to reshape the CPU+GPU market, making AMD almost besieged, with Intel and NVIDIA teaming up in front, and ASIC leader Broadcom behind, increasing pressure.

And the semiconductor Capex stocks that have been lying down, after Intel receives support from the U.S. government and other aspects, may have the potential to increase capital expenditure again, regardless of whether foundry ultimately succeeds, it needs to buy 'shovels' first, thus benefiting semiconductor equipment companies. As for TSMC, the impact of this collaboration is not significant, but still needs to pay attention to breakthroughs in foundry and process.

Risk Disclosure and Statement of this article: Dolphin Research Disclaimer and General Disclosure