AI has finally waited for Wang Chuanfu.

Not long ago, BYD released a less-than-impressive financial report.

The company is facing profit pressure, and the gross profit margin of its automotive business has been continuously squeezed in the fierce price war. The price war, infrastructure investment for overseas expansion, and high - intensity R & D in intelligentization are all eroding profits. In Q1, the net profit for the single quarter showed a rare sequential decline, and the cash flow from operating activities also significantly narrowed due to the phased heavy - asset expansion.

The capital market has started to re - ask a question that has been around for a long time: Has BYD reached its growth ceiling?

After all, the three - pronged approach of electrification, scale - up, and cost control that BYD was most proficient in over the past few years has now become the standard across the entire industry. When all players have pushed blade batteries and the supply chain to the extreme, BYD's original moat is being mercilessly diluted by the industry.

But on May 28th, BYD held a very different press conference.

Wang Chuanfu, the chairman of BYD, stood on the stage and released the Xuanji A3, China's first self - developed 4nm high - end intelligent driving chip. The total computing power of the three chips exceeds 2100 TOPS. This is not just a story on a PPT. The chips have entered the vehicle - installation stage and will be gradually installed in vehicles in the second half of this year.

He said a sentence that is easily overlooked by the industry: A 4nm automotive - grade chip is equivalent to a 2nm chip in consumer electronics.

Then came another set of numbers:

There is a chip R & D team of more than 7,000 people. The cumulative investment in chips has exceeded 100 billion yuan. There are four chip R & D bases and five wafer fabs. From product definition, architecture design, circuit design to wafer manufacturing, packaging, and testing, all seven links are done in - house.

In Wang Chuanfu's words, BYD is the only global automaker with the full - process and full - link manufacturing capabilities for chips.

The real intention behind this statement is highly aggressive.

Even a powerful company like Tesla can only design its FSD chips by itself and must highly rely on TSMC for tape - out. NIO's Shenji chip also cannot get rid of contract manufacturing, and Huawei has also been restricted by advanced manufacturing processes.

Globally, no other automaker can be like BYD, which not only holds the physical core of new energy (battery and IGBT wafer fabs) but also disrupts the computing power table of intelligentization.

Many people's first reaction after watching the press conference was that BYD has finally been forced to get involved in the intelligent driving competition.

But my real feeling is something else. Wang Chuanfu has finally started to bring BYD into the AI era.

I

Over the past few years, there has been an implicit industrial division of labor in the new energy vehicle industry.

The market assumes that intelligentization belongs to the ecological system led by Huawei, AI belongs to Tesla's Silicon Valley narrative, chips belong to NVIDIA's hardware monopoly, and the operating system belongs to the user experience of new car - making forces. BYD, on the other hand, is more like an industrial manufacturing champion in the electrification era.

Its strongest barriers are not intelligentization, not the premium of high - end brands, and not even software capabilities. Instead, it is the ultimate manufacturing efficiency, an airtight supply chain, and the ability to completely reduce the cost of electric vehicles.

So in recent years, BYD has been in a rather delicate position.

On the one hand, it has almost rewritten the global new energy vehicle landscape with its huge throughput. On the other hand, industry elites have always had an arrogant prejudice against it. They admit that BYD can make cars but do not think it can represent the future.

Especially after AI began to redefine the foundation of cars, this kind of voice has become more and more harsh. Many people think that what really matters in future cars is no longer batteries and metal shells, but intelligent driving algorithms, chips, systems, and data closed - loops.

And BYD does not have an advantage in these aspects.

An even more heart - wrenching scene occurred at the investor communication meeting in March 2023.

Wang Chuanfu said a paragraph on the stage that was later widely spread across the internet: Autonomous driving is all nonsense. Creating something flashy is just a deception. It's like the Emperor's New Clothes.

After that statement, he became the target of ridicule in the tech circle. Yu Chengdong publicly mocked his lack of understanding of the industry, and He Xiaopeng directly said that autonomous driving is not nonsense. For a year or two, the backwardness of BYD's intelligentization almost became an industry consensus.

Wang Chuanfu did not make any public response during that time.

So the really interesting thing about this press conference is not the 4nm chip itself, but that Wang Chuanfu, for the first time, clearly showed his cards to the outside world with the most concrete hardware stack. BYD no longer wants to be just a manufacturing champion in the new energy era.

It wants to enter the next round of the game.

II

This event is in line with a major shift that has occurred in the Chinese technology industry over the past few years.

As the AI era has progressed, more and more surviving Chinese companies have started to focus on building underlying heavy - infrastructure capabilities. Huawei has re - emphasized Ascend computing power and the pure - blooded HarmonyOS, Xiaomi has heavily invested in underlying chip R & D, and ByteDance has been frantically building AI infrastructure.

The industry trend has completely shifted from the light - weight and business - model innovation advocated in the past mobile internet era to a more brutal and heavy - asset technology - heavy - industry competition.

But BYD's uniqueness lies in that while others are expanding on their existing AI capabilities, BYD is transitioning from an electrification king to an intelligentization player.

This is a different level of difficulty.

Electrification and intelligentization are two completely different capability models.

Over the past decade, BYD has built a set of physical defenses belonging to the industrial era, including blade batteries, DM - i technology, and a super supply chain. These have made it the world's largest new energy vehicle manufacturer.

But today, the industry is entering the next information stage.

Cars are becoming more and more like AI terminals. Consumers are more concerned about how smart the intelligent driving system is, whether the in - car system will freeze, whether the AI interaction is like that of a real person, and whether the system can be continuously updated via OTA.

In the past, when buying luxury cars, people compared engines, chassis, and mechanical qualities. Today, more and more users are comparing whose system is more like a "next - generation product".

This is exactly what BYD has been least good at in the past.

So although this press conference shows a strong sense of catching up, this catch - up shows that Wang Chuanfu is extremely aware that the real fatal danger in the AI era is never the unappealing short - term profit, but the loss of the right to speak in the final outcome of the next - generation cars.

BYD's greatest success in the past few years was defining how to build electric vehicles. Today, it is starting to compete for something else: who gets to set the underlying standards for AI cars.

This is why the most important protagonist of this press conference is not the new car, but the 4nm chip. What the chip really represents is that BYD is starting to compete for the "brain" of the next - generation cars.

III

To understand Wang Chuanfu's current actions, we need to go back to 2008.

That year, the financial crisis swept through, and BYD was under cash - flow pressure. On October 6th, a few days after Warren Buffett invested in BYD, Wang Chuanfu made a decision that stunned the entire industry. He spent 171 million yuan to take over the bankrupt Ningbo Zhongwei Semiconductor Wafer Factory.

The board of directors almost unanimously voted against it, and even the in - house chip engineers were not optimistic. The industry's evaluation was that Wang Chuanfu was acting in desperation.

At that time, no one understood why a company that made batteries and did contract manufacturing would want to get involved in a wafer factory, which was a cash - flow - consuming mess.

But Wang Chuanfu was not betting on the current market. He believed that the real deciding factor for electric vehicles 15 years later would be power semiconductors. The most critical, highest - tech, and most profitable chip in an electric vehicle is called IGBT.

The return on that bold bet took more than a decade to materialize.

In 2009, BYD produced its first IGBT; in 2018, it released IGBT 4.0. Later, when the global automotive industry was hit by an unprecedented chip shortage and the entire industry was in panic over production halts, only BYD remained unscathed thanks to its self - produced chips.

The subsequent glories, such as the blade battery in 2020 and becoming the global sales champion in 2022, can all be traced back to that autumn in 2008 when no one understood the acquisition of the bankrupt wafer factory.

In 2026, the business structure is very similar to that in 2008.

At the most difficult time when profits are most questioned, all the chips are put back on the table to focus on an underlying direction that the market does not currently recognize or understand. The difference is that the stakes have been upgraded this time.

In 2008, the bet was on IGBT and batteries, which belong to the physical layer of electrification. In 2026, the bet is on intelligent driving chips and full - process wafer manufacturing, which belong to the computing power layer of intelligentization.

More importantly, the Xuanji A3 did not emerge overnight. Based on the R & D cycle of automotive - grade 4nm chips, it was at least in preparation for 3 years. That is to say, during the two years from 2023 to 2024 when everyone was mocking Wang Chuanfu for not understanding intelligentization, he was actually working on this.

He didn't explain; he just quietly built up his cards.

IV

There is an interesting shift in the path here.

Over the past few years, the industry has always thought that BYD's greatest advantage is speed. It has fast product iteration, a fast supply chain, fast price cuts, and fast scale expansion.

But today, Wang Chuanfu is no longer like an efficiency - oriented entrepreneur. He is more like a long - termist entrepreneur.

A chip team of 7,000 people, an investment of 100 billion yuan, and self - built power device wafer fabs are not for improving the quarterly profit of next month. On the contrary, they are dragging down the current financial statements.

In the first quarter of this year, BYD's single - quarter R & D expenses reached as high as 1.13 billion yuan, 720 million yuan higher than the net profit of that quarter.

This is why BYD's financial report is starting to look less good. All real underlying capacity building will eat into profits in the early stage.

This is in line with the stories of ByteDance, Xiaomi, and Kuaishou. More and more companies in China's hard - tech industry are starting to accept that in the AI era, the most expensive asset is no longer internet traffic, but underlying hard - infrastructure. This also means long - term high - altitude losses and long - term lack of understanding from the capital market.

But if you don't do it, you won't even have the qualification to sit at the gaming table. This is what makes Wang Chuanfu most worthy of re - evaluation today.

V

Over the past many years, the entrepreneurs really respected in the Chinese technology industry have been of a different type: light, fast, internet - based, platform - oriented, and traffic - driven.

Wang Chuanfu has always seemed a bit old - fashioned. He doesn't tell sexy stories about the future of humanity like Elon Musk, isn't good at capturing public sentiment like Lei Jun, and doesn't have the natural spotlight - attracting ability like Yu Chengdong.

Most of the time, Wang Chuanfu is so low - key that he is like a super factory director who only buries his head in the workshop.

But today, as AI really enters the deep - water area of the industry, the entire business world suddenly realizes that many of the most core competitions will ultimately converge on the most basic capabilities: chips, underlying systems, manufacturing processes, and supply - chain control.

These things are slow, heavy, and it's hard to tell a high - P/E - ratio story in the capital market. But it is often these things that determine a company's life - and - death barriers in the long run.

More subtly, the global technology industry is experiencing an unprecedented heavy - asset shift.

OpenAI is seeking trillions of funds to self - develop chips and data centers. Google's TPU has been iterated to the seventh generation. Tesla is building a heavy - duty closed - loop of designing its own AI chips and self - building Terafab wafer fabs. The ultimate winners in the AI era almost all seem to be moving towards the heaviest path of vertical integration.

AI has extremely high requirements for the coordination of hardware and systems. Any dependence on external suppliers in terms of latency, bandwidth, or power consumption will become a fatal shortcoming at a critical moment. The traditional light - asset division - of - labor model of finding the best suppliers globally for assembly is facing a systematic collapse in the AI scenario.

Wang Chuanfu's judgment in 2008 that electric vehicles need extreme vertical integration was considered counter - intuitive at that time but has now become an industry consensus.

Wang Chuanfu's judgment in 2026 that intelligent vehicles must be vertically integrated to the core computing power and chip closed - loop is still a bold bet by a minority today. What about three years from now?

This is the most interesting part. As the AI era evolves deeper, entrepreneurs like Wang Chuanfu with heavy - industry system thinking are becoming more and more irreplaceable.

BYD's greatest success in the past was seizing the era's dividend of new energy. But what this press conference really reveals is that Wang Chuanfu definitely doesn't want to rest on the laurels of the old era.

He wants to move on to the next era.

The next era is no longer just the era of electric vehicles. It is the era of the automotive information industry completely reconstructed by AI.

[Beyond the Page] Words:

While writing this, I couldn't help but look through the minutes of the 2023 investor communication meeting again.

When Wang Chuanfu said that autonomous driving was all nonsense, he was actually very calm. It didn't seem like anger or a counter - attack. It was more like someone who already knew where the industry was going.

That year, the entire automotive industry was desperately telling stories about intelligentization.

BYD didn't explain; it just made a 4nm chip.

This article is from the WeChat official account "Beyond the Page", written by Huahua and Banjun, and is published by 36Kr with authorization.