What are AI PCs that Nvidia's Jensen Huang is betting on?

[Daniel Nenni]

June 1 (Reuters) - Nvidia has turned the spotlight on AI PCs after CEO Jensen Huang launched a new chip that would put artificial intelligence capabilities directly into laptops and desktop computers, amid mixed demand for such devices already in the market.

HP (HPQ.N), opens new tab last week said AI-optimized computers helped lift its quarterly results, whereas Dell (DELL.N), opens new tab said in January that the AI boom had not generated the kind of demand ‌it had anticipated.

Here's everything we know about AI PCs:

WHAT DOES "AI PC" MEAN?
Manufacturers say AI PCs can process data more swiftly than traditional ones and can handle a greater volume of AI tasks directly on the device, including chatbots.
They do not have to rely on cloud data centers powering most AI applications like OpenAI's ChatGPT and Anthropic's Claude, and some variants can also support training AI models — a compute-intensive task typically done on servers — locally on the device.

The rise of AI agents, or software that can perform tasks on the computer by ⁠itself with minimal human intervention, is also drawing fresh attention to AI PCs.

Nvidia's (NVDA.O), opens new tab RTX Spark, unveiled ahead of the Computex conference in Taiwan, is part of what the company called an effort with Microsoft (MSFT.O), opens new tab to "reinvent the PC" for the AI era. The new chip has been developed in collaboration with MediaTek (2454.TW), opens new tab to run agents locally rather than relying on cloud computing.

PC makers are hoping such powerful AI features will help draw in buyers as more people lean on generative AI for everything from sending emails to planning vacations.
HP said late in May that AI PCs made up 44% of its PC shipments in the second quarter, up from more than 35% in the previous quarter, helping it top revenue and profit estimates.

However, AI PC adoption could be hampered by the memory chip supply squeeze and rising costs.
Market research firm IDC expects total global PC shipments to decline in 2026 due ‌to memory ⁠shortages, rising component prices and supply constraints, even as higher average selling prices lift market value.

The video player is currently playing an ad. You can skip the ad in 5 sec with a mouse or keyboard

WHAT TECHNOLOGY IS USED IN AI PCS?
AI PCs come with specialized processors called neural processing units that handle the majority of on-device AI workloads.
These NPUs work in tandem with central processing units and graphics processors to manage complex tasks, deliver enhanced processing speeds and power applications such as AI assistants.

WHAT ARE SOME OF THE AI PCS AVAILABLE ON THE MARKET?
Nvidia said RTX Spark laptops and compact desktops ⁠are expected this fall from ASUS (2357.TW), opens new tab, Dell, HP, Lenovo (0992.HK), opens new tab, Microsoft and MSI (2377.TW), opens new tab, with Acer (2353.TW), opens new tab and Gigabyte (2376.TW), opens new tab to follow.

Several of these brands, along with Microsoft and Qualcomm (QCOM.O), opens new tab, already offer Copilot+ PCs, which require processors designed specifically to handle AI tasks on the device.

ARE THERE ANY CONCERNS?
When announced in 2024, Microsoft's "recall" feature had raised some privacy concerns. ⁠The feature would track every action performed on the laptop from voice chats to web browsing, and create a detailed history stored on the device. The user can then search this repository and go through past actions.

Following a strong backlash over privacy and security, Microsoft delayed the release of ⁠the feature and instead made it available through a preview mode for certain users after adding stronger protections. The optional feature is available in the newer Copilot+ PCs.

On the other hand, some experts maintain that managing more AI-related tasks directly on the device offers greater privacy, by eliminating the need to use personal data to train large AI models.

https://www.reuters.com/world/asia-pacific/what-are-ai-pcs-that-nvidias-jensen-huang-is-betting-2026-06-01/

 

[Xebec]

I'm honestly curious what these are going to do that the Co-Pilot+ PCs can't already do..

 

[KevinK]

I'm honestly curious what these are going to do that the Co-Pilot+ PCs can't already do..

I'm guessing that these chips can run far more capable local models and agents than Co-Pilot, but in Windows environment. Up to 128GB of shared memory.

 

[floppydisk]

Apple's hardware has become very popular for running local models because of the unified memory.

 

[FIvr]

I'm honestly curious what these are going to do that the Co-Pilot+ PCs can't already do..

Let's be honest, you know no more about the ongoing AI transformation than being victimized by micro-soft marketing, which is what Co-Pilot truly is.

 

[hist78]

View attachment 4679

June 1 (Reuters) - Nvidia has turned the spotlight on AI PCs after CEO Jensen Huang launched a new chip that would put artificial intelligence capabilities directly into laptops and desktop computers, amid mixed demand for such devices already in the market.

HP (HPQ.N), opens new tab last week said AI-optimized computers helped lift its quarterly results, whereas Dell (DELL.N), opens new tab said in January that the AI boom had not generated the kind of demand ‌it had anticipated.

Here's everything we know about AI PCs:

WHAT DOES "AI PC" MEAN?
Manufacturers say AI PCs can process data more swiftly than traditional ones and can handle a greater volume of AI tasks directly on the device, including chatbots.
They do not have to rely on cloud data centers powering most AI applications like OpenAI's ChatGPT and Anthropic's Claude, and some variants can also support training AI models — a compute-intensive task typically done on servers — locally on the device.

The rise of AI agents, or software that can perform tasks on the computer by ⁠itself with minimal human intervention, is also drawing fresh attention to AI PCs.

Nvidia's (NVDA.O), opens new tab RTX Spark, unveiled ahead of the Computex conference in Taiwan, is part of what the company called an effort with Microsoft (MSFT.O), opens new tab to "reinvent the PC" for the AI era. The new chip has been developed in collaboration with MediaTek (2454.TW), opens new tab to run agents locally rather than relying on cloud computing.

PC makers are hoping such powerful AI features will help draw in buyers as more people lean on generative AI for everything from sending emails to planning vacations.
HP said late in May that AI PCs made up 44% of its PC shipments in the second quarter, up from more than 35% in the previous quarter, helping it top revenue and profit estimates.

However, AI PC adoption could be hampered by the memory chip supply squeeze and rising costs.
Market research firm IDC expects total global PC shipments to decline in 2026 due ‌to memory ⁠shortages, rising component prices and supply constraints, even as higher average selling prices lift market value.

The video player is currently playing an ad. You can skip the ad in 5 sec with a mouse or keyboard

WHAT TECHNOLOGY IS USED IN AI PCS?
AI PCs come with specialized processors called neural processing units that handle the majority of on-device AI workloads.
These NPUs work in tandem with central processing units and graphics processors to manage complex tasks, deliver enhanced processing speeds and power applications such as AI assistants.

WHAT ARE SOME OF THE AI PCS AVAILABLE ON THE MARKET?
Nvidia said RTX Spark laptops and compact desktops ⁠are expected this fall from ASUS (2357.TW), opens new tab, Dell, HP, Lenovo (0992.HK), opens new tab, Microsoft and MSI (2377.TW), opens new tab, with Acer (2353.TW), opens new tab and Gigabyte (2376.TW), opens new tab to follow.

Several of these brands, along with Microsoft and Qualcomm (QCOM.O), opens new tab, already offer Copilot+ PCs, which require processors designed specifically to handle AI tasks on the device.

ARE THERE ANY CONCERNS?
When announced in 2024, Microsoft's "recall" feature had raised some privacy concerns. ⁠The feature would track every action performed on the laptop from voice chats to web browsing, and create a detailed history stored on the device. The user can then search this repository and go through past actions.

Following a strong backlash over privacy and security, Microsoft delayed the release of ⁠the feature and instead made it available through a preview mode for certain users after adding stronger protections. The optional feature is available in the newer Copilot+ PCs.

On the other hand, some experts maintain that managing more AI-related tasks directly on the device offers greater privacy, by eliminating the need to use personal data to train large AI models.

https://www.reuters.com/world/asia-pacific/what-are-ai-pcs-that-nvidias-jensen-huang-is-betting-2026-06-01/

Nvidia and MediaTek are trying to strengthen themselves in the end-to-end AI revolution. If we look at some of the heavyweight players, many of them already have end-to-end ecosystems.

Apple: From servers, notebooks, desktops, smartphones, tablets, smartwatches, headphones, and smart home speakers/hubs to the in-house processors that power them, the A-series and M-series chips. And let's not forget Apple's extensive content distribution, subscription services, and cloud services.

Amazon: Although Amazon does not offer smartphones, it operates the world's largest cloud services business, powered by its own CPUs and accelerators, along with processors from Intel, AMD, and Nvidia. In addition to its various content and subscription services and its massive global e-commerce and delivery network, Amazon also has more than 600 million Alexa-related edge devices worldwide, including Ring, Echo, and Fire products.

Google: Similar to Apple and Amazon, Google has its own CPUs and accelerators, and even more importantly, the Google Search engine and YouTube platform. It also has the Google Nest ecosystem of smart home devices, including door locks, doorbells, cameras, lights, and smart switches, as well as Pixel phones, and the Android ecosystem.

Microsoft: Microsoft is pursuing a similar strategy with in-house developed CPUs and xPUs, although with different areas of strength and emphasis. Its ecosystem includes Azure cloud services, the Windows operating system (client and server), a broad server and client applications, enterprise productivity software, developer tools, gaming platforms, and strong of AI services.

From an AI PC perspective, Nvidia's new client CPU-GPU platform may appear to be a me-too product. However, from the standpoint of building an end-to-end AI platform, Nvidia needs it in order to define and control its own destiny.

 

[Barnsley]

Let's be honest, you know no more about the ongoing AI transformation than being victimized by micro-soft marketing, which is what Co-Pilot truly is.

Microsoft really do hate their customers!!!

Looks like AI not giving the returns expected , so now is time to screw with your customers.

To be honest they are all at it , massive.spend on data centers , how to get that money back

 

[Barnsley]

Nvidia and MediaTek are trying to strengthen themselves in the end-to-end AI revolution. If we look at some of the heavyweight players, many of them already have end-to-end ecosystems.

Apple: From servers, notebooks, desktops, smartphones, tablets, smartwatches, headphones, and smart home speakers/hubs to the in-house processors that power them, the A-series and M-series chips. And let's not forget Apple's extensive content distribution, subscription services, and cloud services.

Amazon: Although Amazon does not offer smartphones, it operates the world's largest cloud services business, powered by its own CPUs and accelerators, along with processors from Intel, AMD, and Nvidia. In addition to its various content and subscription services and its massive global e-commerce and delivery network, Amazon also has more than 600 million Alexa-related edge devices worldwide, including Ring, Echo, and Fire products.

Google: Similar to Apple and Amazon, Google has its own CPUs and accelerators, and even more importantly, the Google Search engine and YouTube platform. It also has the Google Nest ecosystem of smart home devices, including door locks, doorbells, cameras, lights, and smart switches, as well as Pixel phones, smartwatches, and the Android ecosystem.

Microsoft: Microsoft is pursuing a similar strategy with in-house developed CPUs and xPUs, although with different areas of strength and emphasis. Its ecosystem includes Azure cloud services, the Windows operating system (client and server), a broad server and client applications, enterprise productivity software, developer tools, gaming platforms, and strong of AI services.

From an AI PC perspective, Nvidia's new client CPU-GPU platform may appear to be a me-too product. However, from the standpoint of building an end-to-end AI platform, Nvidia needs it in order to define and control its own destiny.

None of this will be good for consumer.

Bring it on I say.

Folk think prices are high now and services are terrible , just wait!!!

 

[KevinK]

Apple's hardware has become very popular for running local models because of the unified memory

My view is that NVIDIA / Microsoft is trying to offer the same on Windows platform.

 

[hist78]

None of this will be good for consumer.

Bring it on I say.

Folk think prices are high now and services are terrible , just wait!!!

Isn't it good by adding Nvidia-MediaTek to the processor supplier list for the client computing, in addition to Intel, AMD, and Qualcomm? More choices and more competition?

 

[Barnsley]

Isn't it good by adding Nvidia-MediaTek to the processor supplier list for the client computing, in addition to Intel, AMD, and Qualcomm? More choices and more competition?

If costs to the end user/ consumer do not come down , then how is said competition helping the masses?

 

[hist78]

If costs to the end user/ consumer do not come down , then how is said competition helping the masses?

Are you thinking more client CPU suppliers will cause higher price and less competition?

 

[Barnsley]

Are you thinking more client CPU suppliers will cause higher price and less competition?

I am thinking that any excuse to put up prices.

Has there been any downward pressure on prices with regards "competition" that you have seen?


As part of the chain that isnt benefiting from higher end prices , it is very very challenging.

The Foundries are seemingly able to jack up prices on a whim, whilst simultaneouslty piling pressure on their suppliers to reduce prices.

Sadly the semicon ecosystem still has not worked out how to be a benefit for all.

 

[Jozo035]

View attachment 4681



Microsoft really do hate their customers!!!

Looks like AI not giving the returns expected , so now is time to screw with your customers.

To be honest they are all at it , massive.spend on data centers , how to get that money back

Reporting is bit misleading and click-baity. Github Copilot was never about heavy agentic coding use.

They gave You 21 premium requests on Gemini 3.5 flash or 33 requests on Sonnet models per month!!! Not even on highest effort.

Literally if You used free version of Claude You got more usage than $10 plan.

 

[Xebec]

I'm guessing that these chips can run far more capable local models and agents than Co-Pilot, but in Windows environment. Up to 128GB of shared memory.

I agree on paper they can run more capable models -- but in terms of end-user experience, what are they actually going to get out of this or do?

I know there's always some high end users that will have a specific use case - but for general population. Will it write MUCH better emails than say a Ryzen APU or Panther Lake laptop, or do create something graphically that they can't do? I'm leaning towards no..

 

[Paul2]

WHAT DOES "AI PC" MEAN?

What does AI even mean by this point? Wall St. is still convinced of that people would paying few hundred dollars for a photoshop filter.

My access to market research data says the exactly opposite. The market has again moved to "big cheap laptop," while the post-covid trend for something more original, and premium is stagnating.

MTK was always a mass market player, and they ate the smartphone market from the bottom, while being one of the last entrants, and having relatively weak SoC expertise.

While the top brand laptops with their chips are not what people sold as MTK based ARM notebooks 15 years ago, they will more or less follow the same scenario, with just a bit more of direct sales to top buyers, and a bit more of top bins going to premium branded models, that will be going to such buyers.

The "AI" wording there is marketing to lure brands, while they firmly set their eyes to capture the mass market without looking much into whatever AI games Western companies want to play.

Transsion smartphones outsell Apple big time. The number of computer users in the second world now exceeds, the first world, and the number in the third world exceeds the second.

Why Lenovo outsold TW laptop makers? Because huge presence, and aggressive, volume pushing sales tactic in the 2nd, and 3rd world, and nothing else.

 

[hist78]

I agree on paper they can run more capable models -- but in terms of end-user experience, what are they actually going to get out of this or do?

I know there's always some high end users that will have a specific use case - but for general population. Will it write MUCH better emails than say a Ryzen APU or Panther Lake laptop, or do create something graphically that they can't do? I'm leaning towards no..

Nvidia and MediaTek are looking at the long term. To extend my previous post about their intention to build an end-to-end ecosystem, here are my thoughts:

Google has cloud services, AI services such as Gemini, server and client applications, client and edge devices (smartphones, smartwatches, and smart home devices), and operating systems (Android and ChromeOS).

Apple is in a very similar position, owning a complete and robust ecosystem. Nvidia and MediaTek, however, participate in only a portion of the value chain compared to Apple or Google. This is a significant challenge for both companies. If the AI evolution continues, they could potentially be squeezed by Apple or Google, especially since they do not control either an operating system or a client/edge platform.

Microsoft is also engaged in fierce competition with Apple and Google. Much of Microsoft's client related business relies on Intel's and AMD's based hardware. This presents a challenge because Intel has been struggling, while AMD remains much smaller than Apple or Google in terms of ecosystem reach. The Nvidia-MediaTek-Microsoft partnership presents an opportunity for the three companies to leverage each other's strengths and address their respective weaknesses. They are effectively building a non-x86 path as a hedge in case Intel cannot catch up, helping ensure that they are not pushed aside by Apple or Google.

The new Nvidia RTX Spark N1 and N1X are obviously too expensive and unnecessary for most mainstream PC users. However, in terms of building a native developer network (rather than relying on x86 translation) and penetrating client environments, they represent a measured starting point. Starting without mass market volume can be a disadvantage, but it also allows Nvidia and MediaTek to cultivate their own market within a smaller and more controllable audience, such as edge AI and client AI developers.

Gaining developers' support is the first step toward building Nvidia's long term client hardware ecosystem. Developers come first then mass market adoption follows.

 

[Xebec]

Gaining developers' support is the first step toward building Nvidia's long term client hardware ecosystem. Developers come first then mass market adoption follows.

This is a well thought out response, and I appreciate it / understand.

It just seems like another 'build it and they will come' scneario, which seems like a high risk gamble with questionable reward.

Maybe it's also a hedging of the bets incase the AI workloads do start to meaningfully shift towards PC (Intel's original strategy with NPU-on-CPU).

(I was hoping there was some clear scenario with N1X/N1 that would be an immediate win for a decent # of users. )

 

[KevinK]

Will it write MUCH better emails than say a Ryzen APU or Panther Lake laptop, or do create something graphically that they can't do? I'm leaning towards no..

I'm betting that early sales will only be for the high-end, high-memory SKUs for running mid-sized AI models locally, mostly for enterprise developers.

 

[hist78]

I'm betting that early sales will only be for the high-end, high-memory SKUs for running mid-sized AI models locally, mostly for enterprise developers.

$4,000 or $5,000 for an N1 or N1X computer may be too expensive for individual consumers, but for large corporate users, it could be justifiable and affordable because of the potential AI capabilities and productivity gains.

I know some major financial firms are deploying AI extensively. Two friends of mine who work in the financial industry told me that they don't know how they could perform their jobs with the same level of efficiency without AI.