Huawei’s Moon Mode Scandal: The Forgotten 2019 AI Fake That Suddenly Feels Nostalgic — As Huawei Prepares to Power DeepSeek V4

In early April 2026, a quiet but seismic piece of news dropped: DeepSeek’s upcoming V4 model — the next major leap from one of China’s strongest open-weight AI labs — will run entirely on Huawei chips, not Nvidia.

DeepSeek has spent months working directly with Huawei (and Cambricon) to rewrite parts of the model’s code so it performs optimally on Huawei’s Ascend AI processors. Chinese tech giants like Alibaba, ByteDance, and Tencent have already placed massive orders for the new Huawei silicon.

For anyone who uses DeepSeek daily, this is big. The current versions are already absurdly good for everyday work — fast, sharp, and surprisingly uncensored. If V4 keeps that momentum while ditching U.S. hardware entirely, it’s another milestone in China’s push for AI self-sufficiency.

The news sent me down a rabbit hole on Huawei itself. I expected some epic underdog story. What I found was… pretty straightforward. Ren Zhengfei, the founder, is the classic Asian-hustle archetype: ex-military, relentless work ethic, started the company in 1987 with almost nothing, and turned it into a global telecom powerhouse through sheer grind and clever positioning. No superhero origin, no dramatic “eureka” moment — just decades of outworking everyone else.

But near the end of my reading, I hit something I had completely forgotten: a loud, messy AI scandal from 2019 that feels almost quaint today.

---

The Moon Mode That Wasn’t

In 2019 Huawei launched the **P30 Pro** and heavily marketed its “Moon Mode.” The pitch was pure magic: point your phone at the moon, zoom in, and the AI would recognize the object and deliver stunning, crater-sharp photos — no tripod, no telescope needed. Marketing videos showed jaw-dropping detail. “This is what your phone can do now,” the ads screamed.

Then the experiments started.

A Chinese researcher (Wang Yue on Zhihu) began testing the feature. He didn’t photograph the actual moon. He pointed the phone at **blurry white circles**, random light sources, even a glowing lamp or a photo of Mars at the right angle. Every single time, the P30 Pro “recognized” it as the moon and outputted a beautifully detailed lunar image complete with realistic craters and highlights.

It wasn’t just enhancing the photo. It was **overlaying or replacing** parts of the image with pre-stored high-quality moon templates. The final picture was part real capture, part AI-generated stock moon.

Huawei denied it was outright faking photos and insisted the AI was only “enhancing” what the sensor captured. The internet, of course, lost its mind. Tech sites ran headlines like “Huawei Caught Faking Moon Photos.” It became one of the biggest camera controversies of the year.

---

Why It Feels Hilarious in 2026

Looking back, the outrage seems almost cute.

Today, every flagship phone uses heavy AI image processing — denoising, super-resolution, style transfer, object hallucination. Modern computational photography quietly fills in details, removes noise, and “improves” reality in ways that would have been called cheating in 2019. The difference? It’s subtler now, and we’ve all quietly accepted it. We expect our phones to make us look better, our sunsets more dramatic, and our food more appetizing.

The Moon Mode incident forced a brief conversation about “the ethics of AI in photography.” Media outlets wrote solemn think-pieces about “crossing boundaries.” But the real lesson was simpler: once AI can make something look convincingly better, most people don’t actually mind — as long as the result is impressive.

And nowhere has AI image and video tech advanced faster than in China. The same ecosystem that once got roasted for faking the moon is now shipping models that generate hyper-realistic video, edit photos in real time, and power the next generation of consumer hardware.

Alo read:

• Meta’s Muse Spark: A Respectable Step Up That Finally Puts Them Back in the Game
• Services: The New Software – Why the Next $1 Trillion Company Will Look Like a Services Firm
• When Cursor Wiped a User's PC: A Cautionary Tale of AI Overreach
• The Great AI Talent Paradox: Why Everyone is Hiring "AI Engineers" but Nobody Can Find Them

---

Full Circle

So here we are in 2026. The company that once got dragged for over-promising what its AI could do with a phone camera is now supplying the silicon that will run one of the world’s most capable open AI models.

DeepSeek V4 on Huawei Ascend chips isn’t just a technical win — it’s a symbolic one. It shows how far the entire Chinese AI stack has come in just seven years. From “AI that fakes the moon” to “AI that runs on home-grown frontier hardware.”

The old scandal feels like ancient history now. But it’s a perfect time capsule: a reminder of how quickly we moved from being shocked by AI trickery to expecting it everywhere — and how the same players who were once accused of going too far are now the ones quietly building the future.

If you’re curious about the original Moon Mode drama, the old Zhihu threads and 2019 tech articles are still out there. They’re worth a quick nostalgic read — mostly to see how dramatically the Overton window on AI “enhancement” has shifted.

The moon shots may have been fake.  
But the progress? That’s very, very real.