The artificial-intelligence race is moving from enormous data centres into something billions of people carry every day: the smartphone.
Taiwanese chip designer MediaTek has introduced its new Dimensity 9600 Pro, a premium mobile processor manufactured using TSMC's advanced 2-nanometre semiconductor technology.
The company says the processor delivers a major improvement in artificial-intelligence performance, positioning MediaTek to compete more aggressively for the premium smartphone market while manufacturers increasingly build generative AI features directly into devices.
The development matters beyond another annual smartphone-chip upgrade.
As AI becomes a central feature of consumer electronics, the ability to perform sophisticated calculations locally — without constantly sending information to distant cloud servers — could become one of the most important battlegrounds in personal technology.
MediaTek Is Moving to 2-Nanometre Technology
The Dimensity 9600 Pro is manufactured using TSMC's latest 2nm process.
In semiconductor manufacturing, terms such as 2nm describe a generation of production technology rather than the literal dimensions of every transistor.
Moving to a newer process generally allows chipmakers to improve performance and energy efficiency while fitting increasingly complex circuitry into compact devices.
Those advantages are especially valuable in smartphones.
Unlike data centres, phones have limited battery capacity and extremely limited space for cooling.
A mobile processor therefore needs to deliver more computing power without producing excessive heat or draining the battery too quickly.
The New Chip Is Built for Generative AI
One of the Dimensity 9600 Pro's most important components is its neural processing unit.
MediaTek says the new NPU provides a 51% performance improvement for on-device generative AI workloads compared with the previous generation.
That is significant because smartphone companies increasingly want AI applications to operate directly on the device.
A phone might summarise text, edit photographs, translate conversations or assist users without sending every request to an external server.
The faster the local AI processor becomes, the more sophisticated those applications can potentially be.
Why On-Device AI Matters
Today's most powerful generative AI systems usually depend heavily on cloud computing.
A user submits a request.
Information travels to a data centre.
Large processors perform the calculation.
The result returns to the user's device.
That model is extremely powerful, but it has limitations.
It requires connectivity.
It consumes expensive server capacity.
It can introduce latency.
And some users may prefer sensitive information to remain on their own devices.
On-device AI offers a different approach.
Your Phone Could Process More Data Locally
Imagine asking a smartphone to summarise private messages.
If the task can be processed locally, those messages may not need to leave the device simply to generate the summary.
The same principle could apply to photographs, recordings, documents and personal information.
Local processing therefore has potential privacy advantages.
It can also work when internet connectivity is limited.
For smartphone manufacturers, these benefits make powerful AI chips increasingly attractive.
Smartphones Are Becoming Small AI Computers
For years, smartphone competition focused heavily on cameras, displays and processor speed.
AI is changing the priorities.
Future devices may increasingly be judged by how well they can understand context, organise information and perform tasks for users.
A smartphone could identify relevant information across messages and calendars.
It might edit images through natural-language instructions.
It could translate conversations in real time.
It may eventually run more capable personal AI assistants that understand information stored locally on the device.
Those features require significant computing power.
MediaTek Is Challenging the Premium Chip Market
MediaTek has historically been particularly strong across Android smartphones, including many affordable and mid-range devices.
But the company has increasingly pushed into the premium market.
The Dimensity 9600 Pro represents another step in that strategy.
Premium smartphone processors can command higher prices and provide greater margins than chips designed primarily for lower-cost devices.
Success at the top of the market also strengthens a semiconductor company's technology reputation.
Xiaomi, Oppo and Vivo Could Be Important
The new processor is expected to appear in upcoming smartphones from manufacturers including Xiaomi, Oppo and Vivo, according to Reuters.
Those companies collectively sell enormous numbers of devices, particularly across Asian markets.
If flagship models adopt the Dimensity 9600 Pro, MediaTek could gain greater visibility among consumers who traditionally associate premium Android performance with rival chip platforms.
The final experience will depend on more than the processor itself.
Phone manufacturers must integrate the chip with software, cameras, batteries and cooling systems.
But the processor establishes the performance foundation.
MediaTek Also Introduced the Dimensity 9600M
The 9600 Pro is not MediaTek's only new processor.
The company has also introduced the Dimensity 9600M, manufactured using TSMC's 3nm technology and intended to reach a broader section of the high-end smartphone market.
That two-chip approach gives MediaTek greater flexibility.
The Pro model can target the highest-performance devices.
The 9600M can bring newer capabilities to additional premium phones without necessarily requiring the most advanced manufacturing process.
TSMC Is Critical to the Story
MediaTek designs chips but does not operate the enormous advanced fabrication plants required to manufacture them.
That role belongs to companies such as TSMC.
The Taiwanese semiconductor manufacturer has become one of the world's most strategically important technology businesses because it produces advanced processors for numerous global customers.
Building leading-edge semiconductor factories requires enormous capital investment and extraordinary engineering precision.
That makes access to the newest manufacturing processes a major competitive advantage.
Why 2nm Chips Are So Difficult to Produce
Modern semiconductor manufacturing operates at extraordinary scales.
Billions of transistors can be packed into a single processor.
Manufacturing those chips requires sophisticated lithography equipment, advanced materials and exceptionally clean production environments.
Even microscopic contamination can ruin a chip.
As manufacturers move towards newer process generations, the technical difficulty rises.
That is why only a small number of companies are capable of producing the world's most advanced semiconductors at commercial scale.
AI Is Increasing Demand for Advanced Chips Everywhere
The AI boom began largely in data centres.
Companies needed enormous quantities of specialised accelerators to train and operate large AI models.
Now the same trend is spreading into consumer devices.
Smartphones need neural processors.
Laptops increasingly include dedicated AI engines.
Cars require sophisticated computing systems.
Robots need processors capable of interpreting their surroundings.
AI is therefore increasing demand across multiple parts of the semiconductor industry.
The Smartphone AI Race Is Different From the Data-Centre Race
Data centres can consume huge amounts of electricity.
Smartphones cannot.
That changes the engineering challenge.
A mobile chip must perform AI calculations while operating within a small battery-powered device.
Efficiency can therefore be just as important as raw speed.
A processor that is extremely powerful but quickly overheats or drains the battery would offer limited practical value.
Advanced manufacturing processes help chip designers improve that balance.
AI Features Could Work Without the Internet
One of the most interesting possibilities is offline AI.
A traveller could potentially use sophisticated translation without a reliable internet connection.
A user could search personal photographs using natural language while keeping the images on the phone.
Voice transcription could happen locally.
AI-powered editing could operate without sending media to the cloud.
Not every AI task will move onto smartphones. The largest models will continue requiring powerful data centres.
But the dividing line between cloud AI and local AI is shifting.
Local AI Could Reduce Cloud Costs
There is another important business advantage.
Running generative AI in data centres can be expensive.
Every user request consumes computing resources.
If smartphone processors can handle simpler AI tasks locally, technology companies may reduce the amount of server infrastructure needed for those requests.
The cloud could then handle only the most demanding tasks.
That creates a hybrid model.
Smaller calculations happen on the phone.
Larger calculations go to the data centre.
Privacy Could Become a Selling Point
Consumer awareness of data privacy continues to increase.
AI makes the issue even more important because assistants may eventually process extremely personal information.
Messages.
Photos.
Voice recordings.
Schedules.
Documents.
Location history.
If more of that processing occurs locally, smartphone companies may be able to offer stronger privacy guarantees.
That could become an important marketing advantage.
MediaTek Is Expanding Beyond Smartphones
The company's ambitions are also moving beyond mobile devices.
MediaTek is expanding into AI data-centre processors and custom chips, according to Reuters.
Its first AI accelerator for a major US cloud provider is expected to enter mass production in late 2026.
That means MediaTek increasingly wants to participate at both ends of the AI computing market.
It can supply processors inside consumer devices while also pursuing chips used in large-scale computing infrastructure.
Nvidia and Alphabet Have Invested
MediaTek recently raised approximately $3.9 billion through a convertible bond offering, with investments from Nvidia and long-time partner Alphabet.
Those relationships are noteworthy because both companies are deeply involved in the AI industry.
Nvidia dominates much of the market for AI accelerators.
Alphabet operates enormous cloud infrastructure and develops its own AI technologies.
Their participation illustrates how interconnected the semiconductor ecosystem has become.
Smartphone Prices Are Under Pressure
Advanced chips are expensive.
So are high-end displays, cameras, memory and batteries.
As component costs rise, smartphone manufacturers face difficult choices.
They can increase retail prices.
They can absorb some of the additional expense.
Or they can change product specifications.
This makes semiconductor economics increasingly important to consumers, even if most buyers never think directly about the chip manufacturing process.
AI Could Give Consumers a Reason to Upgrade Again
Smartphone replacement cycles have lengthened.
Many modern phones remain perfectly usable for several years.
That creates a challenge for manufacturers looking to convince customers to buy new devices frequently.
AI could provide a new upgrade cycle.
If future phones can perform useful tasks that older hardware simply cannot handle efficiently, consumers may have a stronger reason to replace their devices.
Chipmakers are betting heavily on that possibility.
But Consumers Will Judge the Actual Features
Technology companies frequently promote AI performance using benchmarks and technical specifications.
Consumers ultimately care about something simpler:
What can the phone actually do?
A 51% improvement in AI processing is meaningful only if software developers turn that performance into useful applications.
That puts pressure on the entire ecosystem.
Chipmakers need powerful processors.
Phone companies need good hardware.
Software developers need compelling AI features.
Operating systems need to connect everything together.
The Next Smartphone Battle Is About Intelligence
The smartphone industry has already experienced several major competitive eras.
First came larger touchscreens.
Then better cameras.
Then faster mobile networks.
Now artificial intelligence is emerging as another defining battleground.
The winners may be the companies that make AI useful enough that consumers stop thinking of it as a separate feature.
Instead, intelligence becomes embedded throughout the device.
What Happens Next?
The first smartphones using MediaTek's new processors will provide the real test.
Performance benchmarks will matter.
Battery efficiency will matter.
Thermal performance will matter.
But the biggest question will be whether the new AI capabilities noticeably improve everyday smartphone use.
The Dimensity 9600 Pro's 2nm manufacturing process and claimed 51% improvement in on-device generative AI performance show how quickly mobile hardware is evolving.
The wider significance is even bigger.
For the first phase of the generative AI boom, intelligence lived primarily inside enormous data centres.
The next phase is increasingly moving closer to the user.
And one of the most important AI computers may soon be the one already sitting in your pocket.