Noetra, a domestic AI company backed by 44 Japanese firms including SoftBank, NEC, Sony, and Honda, with roughly ¥1 trillion in support from the Ministry of Economy, Trade and Industry (METI), has President Hirotora Tamba voicing alarm: “If we fail, there’s no coming back.”
On 5ch, commenters pointed out that the key researchers behind GPT and NVIDIA were largely immigrants, joked cynically that METI’s involvement alone spells doom, and voiced pessimism that “even ¥10 trillion wouldn’t help — Japan just doesn’t have enough AI talent.”
Noetra, an artificial intelligence development company backed by 44 domestic firms including SoftBank, NEC, Sony Group, and Honda, with roughly ¥1 trillion in total support from the Ministry of Economy, Trade and Industry and others. It is the core company tasked with developing a domestic foundation model to serve as the “brain” for AI operating in the real world — so-called “physical AI” — including self-driving cars and robots. “This is our last chance. If we fail once, there’s no second try.” So declares Hirotora Tamba, who has taken the post of president and representative director. On why it must be homegrown…
Source: nikkei.com / Read the original article
Is there even an organization or person here with the brains to build that brain? I keep seeing American companies pour tens of trillions of yen into this while they're already miles ahead — and we think this is gonna be fine?
The world's gone this globalized and individualistic — there's no way poor Japan can come from behind in a field where it's already fallen behind.
Companies dumped their employees, so patriotism and company loyalty are dead, and all the talented people are fleeing overseas.
Japan and Japanese companies are left with nothing but useless salary thieves like me lol
Companies dumped their employees, so patriotism and company loyalty are dead, and all the talented people are fleeing overseas.
Japan and Japanese companies are left with nothing but useless salary thieves like me lol
That kind of resolve deserves respect — though I'd feel a lot safer if he just stayed off airplanes.
By the way, even just within OpenAI, a lot of the key developers are Polish — and not even American-educated, they're University of Warsaw grads.
Ilya Sutskever, who built the first GPT —
born in Russia, raised in Israel.
Studied at the University of Toronto under Nobel physics laureate Geoffrey Hinton.
Went to Google together with Geoffrey.
The one who recruited them to Google —
Fei-Fei Li, raised in rural China.
Princeton, neuroscience.
Founded Stanford's Human-Centered AI Lab.
Li went to Google Cloud.
Geoffrey and the others went to Google Brain.
Geoffrey's British too, by the way.
Besides Ilya there's another one, Alex —
Alex is the pioneer who brought AI to PCs using NVIDIA GPUs.
NVIDIA owes where it is today to these three.
None of them are American.
Elon Musk —
South Africa.
Peter Thiel, the godfather of the PayPal Mafia and big boss at Palantir —
born in Germany, raised in South Africa.
Where exactly are the Americans in all this?
born in Russia, raised in Israel.
Studied at the University of Toronto under Nobel physics laureate Geoffrey Hinton.
Went to Google together with Geoffrey.
The one who recruited them to Google —
Fei-Fei Li, raised in rural China.
Princeton, neuroscience.
Founded Stanford's Human-Centered AI Lab.
Li went to Google Cloud.
Geoffrey and the others went to Google Brain.
Geoffrey's British too, by the way.
Besides Ilya there's another one, Alex —
Alex is the pioneer who brought AI to PCs using NVIDIA GPUs.
NVIDIA owes where it is today to these three.
None of them are American.
Elon Musk —
South Africa.
Peter Thiel, the godfather of the PayPal Mafia and big boss at Palantir —
born in Germany, raised in South Africa.
Where exactly are the Americans in all this?
>>8
Ugh, purists are exhausting.
Ugh, purists are exhausting.
I started wondering if universities that don't select students based on math even have a reason to exist, back when I was prepping for entrance exams — and now the gap with the rest of the world can't be hidden anymore.
No way Japan can pull this off —
we can't even produce originality.
we can't even produce originality.
Well, it's just a system for skimming off tax money, so nobody really cares if it succeeds or fails.
Nobody takes responsibility anyway.
Nobody takes responsibility anyway.
The endgame of the bureaucrat's board game is landing a cushy "amakudari" post-retirement post (bureaucrats parachuting into cushy jobs) — 10 days of work a year for tens of millions of yen, collecting a retirement bonus, then rinse and repeat. Unless that whole system changes, of course the country's going to fall apart.
"Ministry of Economy, Trade and Industry and others" —
The moment I saw that, I knew this was doomed.
With METI, the ultimate jinx, attached to it, it's completely over.
The moment I saw that, I knew this was doomed.
With METI, the ultimate jinx, attached to it, it's completely over.
A company with so many captains it might just sail straight into outer space. (riff on the proverb "too many captains steer the ship up a mountain")
Honestly, the right move is probably to use Japanese money to just buy up a few American startups.
Unless they let people use it freely during development and harvest the data from that, this isn't going to work.
Pretty sure we're about 120 laps behind at this point…
They could probably distill their way to the same level pretty quickly, right? (referring to AI "distillation," copying a rival model's know-how into a cheaper one)
No, the question is why they didn't start this sooner.
Japan can't do this, might as well give up already.
Lots of shills in here too.
The moment the investment dries up, it's over.
The moment the investment dries up, it's over.
>>41
If it were the latest version I'd be all for it, but what's the point of funding something that's already behind? Just another vested-interest racket.
If it were the latest version I'd be all for it, but what's the point of funding something that's already behind? Just another vested-interest racket.
Not totally sure, but isn't this already behind the times at this point?
I'd get it if they threw a huge budget at something on par with the latest US/China AI, but
funding something that's already behind — isn't that just wasting tax money?
I'd get it if they threw a huge budget at something on par with the latest US/China AI, but
funding something that's already behind — isn't that just wasting tax money?
Not totally sure, but isn't this already behind the times at this point?
I'd get it if they poured a huge budget into something on par with the latest US/China AI, but
throwing a huge budget at a version that's already behind — isn't that just wasting tax money?
Develop the latest model first, then go all in at once.
I'd get it if they poured a huge budget into something on par with the latest US/China AI, but
throwing a huge budget at a version that's already behind — isn't that just wasting tax money?
Develop the latest model first, then go all in at once.
Can't they just take a page from China's book — rip off everything and ignore human rights while they're at it?
Even in the world of car restoration and customization, the US is already on another level with 3D scanning and CNC machining.
It's probably because the incompetent always say "AI's so amazing I don't need to try anymore." Corporate recruiters should really sharpen their skills at spotting the incompetent people who've just been leaning on AI a little.
Even a useless AI company like Tier IV wins the moment it goes public. (Tier IV: a Japanese self-driving startup)
By the time this thing's finished, all I can picture is a future where we're still laps behind.
Don't get so dramatic over a mere ¥1 trillion.
Building just one frontier LLM probably costs about that much anyway.
And plenty of projects burn that much money and still flop — that's just normal.
Building just one frontier LLM probably costs about that much anyway.
And plenty of projects burn that much money and still flop — that's just normal.
If they're seriously aiming for the next frontier, ¥1 trillion isn't nearly enough.
These guys can't even build an original LLM, and now they're going for physical AI… pfft.
Even ¥10 trillion wouldn't be enough.
There just aren't enough people who can actually develop AI.
Same reason Toyota's behind on self-driving despite having plenty of money.
There just aren't enough people who can actually develop AI.
Same reason Toyota's behind on self-driving despite having plenty of money.
It comes down to the gap in ultra-wealthy backers.
In the end,
of course it's going to fail — this is a one-shot bet that's not allowed to fail.
The strength of the US
is capital that can prop up perpetually money-losing startups,
and the ultra-rich who make that possible.
Even if OpenAI or Anthropic fail, there are endless replacements waiting.
That's depth of bench.
Corporate slaves, bureaucrats, bank employees — none of them can pull this off.
They start infighting immediately,
too busy dodging responsibility and blaming everyone else.
In the end,
of course it's going to fail — this is a one-shot bet that's not allowed to fail.
The strength of the US
is capital that can prop up perpetually money-losing startups,
and the ultra-rich who make that possible.
Even if OpenAI or Anthropic fail, there are endless replacements waiting.
That's depth of bench.
Corporate slaves, bureaucrats, bank employees — none of them can pull this off.
They start infighting immediately,
too busy dodging responsibility and blaming everyone else.
>>56
That "ultra-rich who make that possible" —
that's money from wealthy people worldwide,
not just American money.
That's exactly its biggest strength, and
it's not remotely for the benefit of the American public.
Meanwhile, Japan's tax money and policy aren't for the benefit of its citizens either.
Sorry to the guy in >>1, but on the macro level,
this is nothing more than a sacrificial offering so politicians and bureaucrats can look like they're doing something.
That "ultra-rich who make that possible" —
that's money from wealthy people worldwide,
not just American money.
That's exactly its biggest strength, and
it's not remotely for the benefit of the American public.
Meanwhile, Japan's tax money and policy aren't for the benefit of its citizens either.
Sorry to the guy in >>1, but on the macro level,
this is nothing more than a sacrificial offering so politicians and bureaucrats can look like they're doing something.
Japan was actually ahead in AI frameworks and ASICs, you know.
A Japanese AI company —
Preferred Networks.
Then got overtaken and left in the dust by latecomers.
They had foresight.
They had vision.
So why?
A Japanese AI company —
Preferred Networks.
Then got overtaken and left in the dust by latecomers.
They had foresight.
They had vision.
So why?
*This article is compiled from excerpts and a summary of the 5ch (Business News+) thread [AI] Noetra President Tamba’s Anxiety over Domestic AI Development: “If We Fail, There’s No Coming Back”.
Leave a Reply