wilmott.com

A situation report from the real world:

https://zarazhang.substack.com/p/shadow ... paying-for

Happens all the time, down all the years. CIO, CEO, CFO, CTO are all ballast.

https://en.wikipedia.org/wiki/The_Soul_of_a_New_Machine

Tom West, the leader of the Westborough designers, starts a skunkworks project. Code-named "Eagle", it becomes a backup plan in case Fountainhead fails, and then the company's only hope in catching up with DEC. In order to complete the project on time, West takes risks: he elects to use new technology, and he relies on new college graduates (who have never designed anything so complex) as the bulk of his design team. The book follows many of the designers as they give almost every waking moment of their lives to design and debug the new machine.

That’s rather prescient. It sounds almost exactly like Elon Musk’s program to destroy the US government.

A better analogy is what I was thinking of

DeepSeek's models are "open weight", which provides less freedom for modification than true open-source software.[13][14] The company reportedly recruits AI researchers from top Chinese universities[11] and hires from outside the computer science field to diversify its models' knowledge and abilities.[8]
 
The low cost of training and running the language model was attributed to Chinese firms' lack of access to Nvidia chipsets, which were restricted by the US as part of the ongoing trade war between the two countries. This breakthrough in reducing expenses while increasing efficiency and maintaining the model's performance power and quality in the AI industry sent "shockwaves" through the market. It threatened the dominance of AI leaders like Nvidia and contributed to the largest drop in US stock market history, with Nvidia alone losing $600 billion in market value.[15][16]

necessity is the mother of invention (Plato)

Compare to OpenAI (looks kinda bloated)
OpenAI's rapid expansion and rising influence in artificial intelligence are underscored by its growth from 770 employees in November 2023 to 3,531 individuals as of September 2024.

https://www.wheresyoured.at/power-cut/

Ed Zitron tends to be critical hyperdoom but worth a read all the same. 

Apple's very condensed summary...... how right or wrong it is I'm not going to wager. 

Microsoft has cancelled data centre leases and pulled back on expansion plans, suggesting a lack of confidence in generative AI and OpenAI’s future. This move, equivalent to over 14% of Microsoft’s current capacity, contradicts previous statements about the potential of generative AI. The cancellation of data centre contracts and Microsoft’s decision to allow OpenAI to work with Oracle indicate a shift in Microsoft’s strategy and a potential reevaluation of its investment in OpenAI.

Is it canary 2.0 in the coalmine?

Is it canary 2.0 in the coalmine?

Cheep! 

Not too sure, letters of intent are just that, intent. MSFT have made a bit of noise about things not panning out though. I read only 1% of the users of Office are using the built in GenAI tools. I am not one of them....

Is it canary 2.0 in the coalmine?

Cheep! 

Not too sure, letters of intent are just that, intent. MSFT have made a bit of noise about things not panning out though. I read only 1% of the users of Office are using the built in GenAI tools. I am not one of them....

Office 365 was down last weekend for a while.
Apart from anything else, there won't be enough water to keep the centres cooled.

Alibaba co-founder & chairman Joseph Tsai disagrees with US Big Tech’s insane R&D and CapEx spend on AI:
 
"DeepSeek is significant because they demonstrated that you can use engineering innovation to drastically lower the cost of training and inference for large language models. Up to this point, people have been massively investing in frontier models. But here’s the problem: if you think about training frontier models as educating your kids to be the smartest - acquiring PhDs in multiple subjects like math, physics, biology, and psychology - there are only five or six rich parents who can afford that" - Joseph Tsai
 
He later added:
 
"You have to ask, what is the purpose of developing AI? Is it to have the smartest child who can win every Nobel Prize, or is it something different - like solving real-world problems and making an economic impact? I think people will start shifting away from investing hundreds of billions into compute infrastructure for the smartest AI and instead focus on actual problems in the world and use AI to solve them" - Joseph Tsai
 

For the betting person, capitalism’s imitative herd mentality brings with it a predictable tendency to overdo new things, creating a steady stream of opportunities to make money during both early valuation runups followed by near inevitable rationalizations. Internet hype and what I see as increasing social impatience (why else are people still paying up for seaside property that will in a handful or two of years will be underwater?) has amplified this tendency. AI will surely settle down in due course into a smaller-scaled and increasingly useful activity, having left in its wake a good deal of zero-sum wealth transfer created during its birthing.

AI will surely settle down in due course into a smaller-scaled and increasingly useful activity

That's a moot question.

It is known what these useful products will be, beyond assistants for Office 365 or GO world champion?
$80 bn investments are mind-boggling.

Cheap at half the price.

Cheap at half the price.

At what point do you want me to chime in with, "I told you so"....... the losses are stacking up and the hype is dying down. 
I mean, it's not like I wrote about it: https://wilmott.com/goldman-sachs-pours ... ai-future/

Though Goldman seem to roll back on their view really quickly for some reason, I still would stick by the report. 

At what point do you want me to chime in with, "I told you so"....... the losses are stacking up and the hype is dying down. 
I mean, it's not like I wrote about it: https://wilmott.com/goldman-sachs-pours ... ai-future/

Though Goldman seem to roll back on their view really quickly for some reason, I still would stick by the report. 

Maybe this is a leading indicator?

“After setbacks, Microsoft’s AI guru Mustafa Suleyman appears to be making slow progress helping the company wean itself off OpenAI” 

Microsoft has poured over $13 billion into the AI firm since 2019, but now it wants more control over its own models and costs. Simple enough in theory—build in-house alternatives, cut expenses, and call the shots. But as a new report from The Information lays out, reality is a lot messier. Or as reporter Aaron Holmes put it, “easier said than done.”
“Last fall, during a video call with senior leaders at OpenAI and Microsoft, Suleyman—who leads Microsoft’s in-house artificial intelligence unit—wanted OpenAI staffers to explain how its latest model, o1, worked, according to someone present for the conversation and two other Microsoft employees who were briefed on it.”

obvious!! remember?

OS/2 v Win
Java v C#
Harvard Graphics v Excel
Word v Wordpervert (a blood bath)
COM v all the other interop stuff
MFC (Microsoft Frustration Classes) v AWT (Awful Windowing Toolkit)
SQL Server v FoxBase

Capische?

OS/2 v Win
Java v C#
Harvard Graphics v Excel
Word v Wordpervert (a blood bath)
COM v all the other interop stuff
MFC (Microsoft Frustration Classes) v AWT (Awful Windowing Toolkit)
SQL Server v FoxBase

Capische?

This is the deleted scene from Pulp Fiction where Mia Wallace is questioning Vincent Vega was related to Suzanne Vega? 
"...an Java man should love it..."

Microsoft developing AI reasoning models to compete with OpenAI, The Information reports

https://finance.yahoo.com/news/microsof ... 1986wx2zzI