Microsoft: Marketing Is One Thing, a Cost Black Hole Is Quite Another
March 11, 2025
Yep, another dinobaby original.
I read “Microsoft Cuts Data Centre Plans and Hikes Prices in Push to Make Users Carry AI Cost.” The headline meant one thing to me: The black hole of AI costs must be capped. For my part, I try to avoid MSFT AI. After testing the Redmoanians’ smart software for months, I decided, “Nope.”
The write up says:
Last week, Microsoft unceremoniously pulled back on some planned data centre leases. The move came after the company increased subscription prices for its flagship 365 software by up to 45%, and quietly released an ad-supported version of some products. The tech giant’s CEO, Satya Nadella, also recently suggested AI has so far not produced much value.
No kidding. I won’t go into the annoyances. AI in Notepad? Yeah, great thinking like that which delivered Bob to users who loved Clippy.
The essay notes:
Having sunk billions into generative AI, Microsoft is trying to find the business model that will make the technology profitable.
Maybe someday, but that day is not today or tomorrow. If anything, Microsoft is struggling with old-timey software as well. The Register, a UK online publication, reports:
Back to AI. The AI financial black hole exists, and it may not be easy to resolve. What’s the fix? Here’s the Microsoft data center plan as of March 2025:
As AI infrastructure costs rise and model development evolves, shifting the costs to consumers becomes an appealing strategy for AI companies. While big enterprises such as government departments and universities may manage these costs, many small businesses and individual consumers may struggle.
Several observations are warranted:
- What happens if Microsoft cannot get consumers to pay the AI bills?
- What happens if people like this old dinobaby don’t want smart software and just shift to work flows without Microsoft products?
- What happens if the marvel of the Tensor and OpenAI’s and others’ implementations continue to hallucinate creating more headaches than the methods improve?
Net net: Marketing may have gotten ahead of reality, but the black hole of costs are very real and not hallucinations. Can Microsoft escape a black hole like this one?
Stephen E Arnold, March 11, 2025
Next-Gen IT Professionals: Up for Doing a Good Job?
March 10, 2025
The entirety of the United States is facing a crisis when it comes to decent paying jobs. Businesses are watching their budgets like misers clutch their purse strings, so they’re hiring the cheapest tech workers possible. Medium explains that “8 Out Of 10 Senior Engineers Feel Undervalued: The Hidden Crisis In Tech’s Obsession With Junior Talent.”
Another term for budgeting and being cheaper is “cost optimization.” Experienced tech workers are being replaced with green newbies who wouldn’t know how to find errors if it was on the back of their hands. Or the experienced tech workers are bogged down by mentoring/fixing the mistakes of their younger associates.
It’s a recipe for disaster, but cost optimization is what businesses care about. There will be casualties in the trend, not all of them human:
“The silent casualties of this trend:
1. Systems designed by juniors who’ve never seen a server catch fire
2. Codebases that work right up until they don’t
3. The quiet exodus of graybeards into early retirement”
Junior tech workers are cheaper, but it is difficult to just ask smart software to impart experience in a couple hundred words. Businesses are also treating their seasoned employees like they are mentors:
“I’m all for mentoring. But when companies treat seniors as:
- Free coding bootcamp instructors
- Human linters for junior code
- On-call explainers of basic algorithms
…they’re not paying for mentorship. They’re subsidizing cheap labor with senior salaries.”
There’s a happy medium where having experienced tech experts work with junior tech associates can be beneficial for those involved. It is cheaper to dump the dinobabies and assume that those old systems can be fixed when they go south.
Whitney Grace, March 10, 2025
AI Generated Code Adds To Technical Debt
March 7, 2025
Technical debt refers to using flawed code that results in more work. It’s okay for projects to be ruled out with some technical debt as long as it is paid back. The problem comes when the code isn’t corrected and it snowballs into a huge problem. LeadDev explores how AI code affects projects: “How AI Generated Code Compounds Technical Debt.” The article highlights that it has never been easier to write code especially with AI, but there’s a large amassment of technical debt. The technical debt is so large that it is comparable to the US’s ballooning debt.
GitClear tracked the an eight-gold increase in code frequency blocks with give or more lines that duplicate adjectives code during 2024. This was ten times higher than the previous two years. GitClear found some more evidence of technical debt:
“That same year, 46% of code changes were new lines, while copy-pasted lines exceeded moved lines. “Moved,” lines is a metric GitClear has devised to track the rearranging of code, an action typically performed to consolidate previous work into reusable modules. “Refactored systems, in general, and moved code in particular, are the signature of code reuse,” says Bill Harding, CEO of Amplenote and GitClear. A year-on-year decline in code movement suggests developers are less likely to reuse previous work, a marked shift from existing industry best practice that would lead to more redundant systems with less consolidation of functions.”
These facts might not seem alarming, especially if one reads Google’s 2024 DORA report that said there was a 25% increase in AI usage to quicken code reviews and documentation. The downside was a 7.2% decrease in delivery and stability. These numbers might be small now but what is happening is like making a copy of a copy of a copy: the integrity is lost.
It’s also like relying entirely on spellcheck to always correct your spelling and grammar. While these are good tools to have, what will you do when you don’t have fundamentals in your toolbox or find yourself in a spontaneous spelling bee?
Whitney Grace, March 7, 2025
Attention, New MBAs in Finance: AI-gony Arrives
March 6, 2025
Another post from the dinobaby. Alas, no smart software used for this essay.
I did a couple of small jobs for a big Wall Street outfit years ago. I went to meetings, listened, and observed. To be frank, I did not do much work. There were three or four young, recent graduates of fancy schools. These individuals were similar to the colleagues I had at the big time consulting firm at which I worked earlier in my career.
Everyone was eager and very concerned that their Excel fevers were in full bloom: Bright eyes, earnest expressions, and a gentle but persistent panting in these meetings. Wall Street and Wall Street like firms in London, England, and Los Angeles, California, were quite similar. These churn outfits and deal makers shared DNA or some type of quantum entanglement.
These “analysts” or “associates” gathered data, pumped it into Excel spreadsheets set up by colleagues or technical specialists. Macros processed the data and spit out tables, charts, and graphs. These were written up as memos, reports for those with big sticks, or senior deciders.
My point is that the “work” was done by cannon fodder from well-known universities business or finance programs.
Well, bad news, future BMW buyers, an outfit called PublicView.ai may have curtailed your dreams of a six figure bonus in January or whatever month is the big momma at your firm. You can take a look at example outputs and sign up free at https://www.publicview.ai/.
If the smart product works as advertised, a category of financial work is going to be reshaped. It is possible that fewer analyst jobs will become available as the gathering and importing are converted to automated workflows. The meetings and the panting will become fewer and father between.
I don’t have data about how many worker bees power the Wall Street type outfits. I showed up, delivered information when queried, departed, and sent a bill for my time and travel. The financial hive and its quietly buzzing drones plugged away 10 or more hours a day, mostly six days a week.
The PublicView.ai FAQ page answers some basic questions; for example, “Can I perform quantitative analysis on the files?” The answer is:
Yes, you can ask Publicview to perform computations on the files using Python code. It can create graphs, charts, tables and more.
This is good news for the newly minted MBAs with programming skills. The bad news is that repeatable questions can be converted to workflows.
Let’s assume this product is good enough. There will be no overnight change in the work for existing employees. But slowly the senior managers will get the bright idea of hiring MBAs with different skills, possibly on a contract basis. Then the work will begin to shift to software. At some point in the not-to-distant future, jobs for humans will be eliminated.
The question is, “How quickly can new hires make themselves into higher value employees in what are the early days of smart software?”
I suggest getting on a fast horse and galloping forward. Donkeys with Excel will fall behind. Software does not require health care, ever increasing inducements, and vacations. What’s interesting is that at some point many “analyst” jobs, not just in finance, will be handled by “good enough” smart software.
Remember a 51 percent win rate from code that does not hang out with a latte will strike some in carpetland as a no brainer. The good news is that MBAs don’t have a graduate degree in 18th century buttons or the Brutalist movement in architecture.
Stephen E Arnold, March 6, 2025
Big Thoughts On How AI Will Affect The Job Market
March 4, 2025
Every time there is an advancement in technology, humans are fearful they won’t make an income. While some jobs disappeared, others emerged and humans adapted to the changes. We’ll continue to adapt as AI becomes more integral in society. How will we handle the changes?
Anthropic, a big player in the OpenAI field, launched The Anthropic Index to understand AI’s effects on labor markers and the economy. Anthropic claims it’s gathering “first-of-its” kind data from Claude.ai anonymized conversations. This data demonstrates how AI is incorporated into the economy. The organization is also building an open source dataset for researchers to use and build on their findings. Anthropic surmises that this data will help develop policy on employment and productivity.
Anthropic reported on their findings in their first paper:
• “Today, usage is concentrated in software development and technical writing tasks. Over one-third of occupations (roughly 36%) see AI use in at least a quarter of their associated tasks, while approximately 4% of occupations use it across three-quarters of their associated tasks.
• AI use leans more toward augmentation (57%), where AI collaborates with and enhances human capabilities, compared to automation (43%), where AI directly performs tasks.
• AI use is more prevalent for tasks associated with mid-to-high wage occupations like computer programmers and data scientists, but is lower for both the lowest- and highest-paid roles. This likely reflects both the limits of current AI capabilities, as well as practical barriers to using the technology.”
The Register put the Anthropic report in layman’s terms in the article, “Only 4 Percent Of Jobs Rely Heavily On AI, With Peak Use In Mid-Wage Roles.” They share that only 4% of jobs rely heavily on AI for their work. These jobs use AI for 75% of their tasks. Overall only 36% of jobs use AI for 25% of their tasks. Most of these jobs are in software engineering, media industries, and educational/library fields. Physical jobs use AI less. Anthropic also found that 57% of these jobs use AI to augment human tasks and 43% automates them.
These numbers make sense based on AI’s advancements and limitations. It’s also common sense that mid-tier wage roles will be affected and not physical or highly skilled labor. The top tier will surf on money; the water molecules are not so lucky.
Whitney Grace, March 4, 2025
AI Research Tool from Perplexity Is Priced to Undercut the Competition
February 26, 2025
Are prices for AI-generated research too darn high? One firm thinks so. In a Temu-type bid to take over the market, reports VentureBeat, "Perplexity Just Made AI Research Crazy Cheap—What that Means for the Industry." CEO Aravind Srinivas credits open source software for making the move possible, opining that "knowledge should be universally accessible." Knowledge, yes. AI research? We are not so sure. Nevertheless, here we are. The write-up describes the difference in pricing:
"While Anthropic and OpenAI charge thousands monthly for their services, Perplexity offers five free queries daily to all users. Pro subscribers pay $20 monthly for 500 daily queries and faster processing — a price point that could force larger AI companies to explain why their services cost up to 100 times more."
Not only is Perplexity’s Deep Research cheaper than the competition, crows the post, its accuracy rivals theirs. We are told:
"[Deep Research] scored 93.9% accuracy on the SimpleQA benchmark and reached 20.5% on Humanity’s Last Exam, outperforming Google’s Gemini Thinking and other leading models. OpenAI’s Deep Research still leads with 26.6% on the same exam, but OpenAI charges $200 percent for that service. Perplexity’s ability to deliver near-enterprise level performance at consumer prices raises important questions about the AI industry’s pricing structure."
Well, okay. Not to stray too far from the point, but is a 20.5% or a 26.6% on Humanity’s Last Exam really something to brag about? Last we checked, those were failing grades. By far. Isn’t it a bit too soon to be outsourcing research to any LLM? But I digress.
We are told the low, low cost Deep Research is bringing AI to the micro-budget masses. And, soon, to the Windows-less—Perplexity is working on versions for iOS, Android, and Mac. Will this spell disaster for the competition?
Cynthia Murrell, February 26, 2025
Rest Easy. AI Will Not Kill STEM Jobs
February 25, 2025
Written by a dinobaby, not smart software. But I would replace myself with AI if I could.
Bob Hope quipped, “A sense of humor is good for you. Have you ever heard of a laughing hyena with heart burn?” No, Bob, I have not.
Here’s a more modern joke for you from the US Bureau of Labor Statistics circa 2025. It is much fresher than Mr. Hope’s quip from a half century ago.
The Bureau of Labor Statistics says:
Employment in the professional, scientific, and technical services sector is forecast to increase by 10.5% from 2023 to 2033, more than double the national average. (Source: Investopedia)
Okay, I wonder what those LinkedIn, XTwitter, and Reddit posts about technology workers not being able to find jobs in these situations:
- Recent college graduates with computer science degrees
- Recently terminated US government workers from agencies like 18F
- Workers over 55 urged to take early retirement?
The item about the rosy job market appeared in Slashdot too. Here’s the quote I noted:
Employment in the professional, scientific, and technical services sector is forecast to increase by 10.5% from 2023 to 2033, more than double the national average. According to the BLS, the impact AI will have on tech-sector employment is highly uncertain. For one, AI is adept at coding and related tasks. But at the same time, as digital systems become more advanced and essential to day-to-day life, more software developers, data managers, and the like are going to be needed to manage those systems. "Although it is always possible that AI-induced productivity improvements will outweigh continued labor demand, there is no clear evidence to support this conjecture," according to BLS researchers.
Robert Half, an employment firm, is equally optimistic. Just a couple of weeks ago, that outfit said:
Companies continue facing strong competition from other firms for tech talent, particularly for candidates with specialized skills. Across industries, AI proficiency tops the list of most-sought capabilities, with organizations needing expertise for everything from chatbots to predictive maintenance systems. Other in-demand skill areas include data science, IT operations and support, cybersecurity and privacy, and technology process automation.
What am I to conclude from these US government data? Here are my preliminary thoughts:
- The big time consulting firms are unlikely to change their methods of cost reduction; that is, if software (smart or dumb) can do a job for less money, that software will be included on a list of options. Given a choice of going out of business or embracing smart software, a significant percentage of consulting firm clients will give AI a whirl. If AI works and the company stays in business or grows, the humans will be repurposed or allowed to find their future elsewhere.
- The top one percent in any discipline will find work. The other 99 percent will need to have family connections, family wealth, or a family business to provide a boost for a great job. What if a person is not in the top one percent of something? Yeah, well, that’s not good for quite a few people.
- The permitted dominance of duopolies or oligopolies in most US business sectors means that some small and mid-sized businesses will have to find ways to generate revenue. My experience in rural Kentucky is that local accounting, legal, and technology companies are experimenting with smart software to boost productivity (the MBA word for cheaper work functions). Local employment options are dwindling because the smaller employers cannot stay in business. Potential employees want more pay than the company can afford. Result? Downward spiral which appears to be accelerating.
Am I confident in statistics related to wages, employment, and the growth of new businesses and industrial sectors? No, I am not. Statistical projects work pretty well in nuclear fuel management. Nested mathematical procedures in smart software work pretty well for some applications. Using smart software to reduce operating costs work pretty well right now.
Net net: Without meaningful work, some of life’s challenges will spark unanticipated outcomes. Exactly what type of stress breaks a social construct? Those in the job hunt will provide numerous test cases, and someone will do an analysis. Will it be correct? Sure, close enough for horseshoes.
Stop complaining. Just laugh as Mr. Hope noted. No heartburn and cost savings too boot.
Stephen E Arnold, February 25, 2025
Are These Googlers Flailing? (Yes, the Word Has “AI” in It Too)
February 12, 2025
Is the Byte write up on the money? I don’t know, but I enjoyed it. Navigate to “Google’s Finances Are in Chaos As the Company Flails at Unpopular AI. Is the Momentum of AI Starting to Wane?” I am not sure that AI is in its waning moment. Deepseek has ignited a fire under some outfits. But I am not going to critic the write up. I want to highlight some of its interesting information. Let’s go, as Anatoly the gym Meister says, just with an Eastern European accent.
Here’s the first statement in the article which caught my attention:
Google’s parent company Alphabet failed to hit sales targets, falling a 0.1 percent short of Wall Street’s revenue expectations — a fraction of a point that’s seen the company’s stock slide almost eight percent today, in its worst performance since October 2023. It’s also a sign of the times: as the New York Times reports, the whiff was due to slower-than-expected growth of its cloud-computing division, which delivers its AI tools to other businesses.
Okay, 0.1 percent is something, but I would have preferred the metaphor of the “flail” word to have been used in the paragraph begs for “flog,” “thrash,” and “whip.”
I used Sam AI-Man’s AI software to produce a good enough image of Googlers flailing. Frankly I don’t think Sam AI-Man’s system understands exactly what I wanted, but close enough for horseshoes in today’s world.
I noted this information and circled it. I love Gouda cheese. How can Google screw up cheese after its misstep with glue and cheese on pizza. Yo, Googlers. Check the cheese references.
Is Alphabet’s latest earnings result the canary in the coal mine? Should the AI industry brace for tougher days ahead as investors become increasingly skeptical of what the tech has to offer? Or are investors concerned over OpenAI’s ChatGPT overtaking Google’s search engine? Illustrating the drama, this week Google appears to have retroactively edited the YouTube video of a Super Bowl ad for its core AI model called Gemini, to remove an extremely obvious error the AI made about the popularity of gouda cheese.
Stalin revised history books. Google changes cheese references for its own advertising. But cheese?
The write up concludes with this, mostly from American high technology watching Guardian newspaper in the UK:
Although it’s still well insulated, Google’s advantages in search hinge on its ubiquity and entrenched consumer behavior,” Emarketer senior analyst Evelyn Mitchell-Wolf told The Guardian. This year “could be the year those advantages meaningfully erode as antitrust enforcement and open-source AI models change the game,” she added. “And Cloud’s disappointing results suggest that AI-powered momentum might be beginning to wane just as Google’s closed model strategy is called into question by Deepseek.”
Does this constitute the use of the word “flail”? Sure, but I like “thrash” a lot. And “wane” is good.
Stephen E Arnold, February 12, 2025
Deepseek: Details Surface Amid Soft Numbers
February 7, 2025
We have smart software, but the dinobaby continues to do what 80 year olds do: Write the old-fashioned human way. We did give up clay tablets for a quill pen. Works okay.
I read “Research exposes Deepseek’s AI Training Cost Is Not $6M, It’s a Staggering $1.3B.” The assertions in the write up are interesting and closer to the actual cost of the Deepseek open source smart software. Let’s take a look at the allegedly accurate and verifiable information. Then I want to point out two costs not included in the estimated cost of Deepseek.
The article explains that the analysis for training was closer to $1.3 billion. I am not sure if this estimate is on the money, but a higher cost is certainly understandable based on the money burning activities of outfits like Microsoft, OpenAI, Facebook / Meta, and the Google, among others.
The article says:
In its latest report, SemiAnalysis, an independent research company, has spotlighted Deepseek, a rising player in the AI landscape. The SemiAnalysis challenges some of the prevailing narratives surrounding Deepseek’s costs and compares them to competing technologies in the market. One of the most prominent claims in circulation is that Deepseek V3 incurs a training cost of around $6 million.
One important point is that building and making available for free a smart software system incurs many costs. The consulting firm has narrowed its focus to training costs.
The write up reports:
The $6 million estimate primarily considers GPU pre-training expenses, neglecting the significant investments in research and development, infrastructure, and other essential costs accruing to the company. The report highlights that Deepseek’s total server capital expenditure (CapEx) amounts to an astonishing $1.3 billion. Much of this financial commitment is directed toward operating and maintaining its extensive GPU clusters, the backbone of its computational power.
But “astonishing.” Nope. Sam AI-Man tossed around numbers in the trillions. I am not sure we will ever know how much Amazon, Facebook, Google, and Microsoft — to name four outfits — have spent in the push to win the AI war, get a new monopoly, and control everything from baby cams to zebra protection in South Africa.
I do agree that the low ball number was low, but I think the pitch for this low ball was a tactic designed to see what a Chinese-backed AI product could do to the US financial markets.
There are some costs that neither the SemiAnalytics outfit or the Interesting Engineering wordsmith considered.
First, if you take a look at the authors of the Deepseek ArXiv papers you will see a lot of names. Most of these individuals are affiliated with Chinese universities. How we these costs handled? My hunch is that the costs were paid by the Chinese government and the authors of the paper did what was necessary to figure out how to come up with a “do more for less” system. The idea is that China, hampered by US export restrictions, is better at AI than the mythological Silicon Valley. Okay, that’s a good intelligence operation: Test destabilization with a reasonably believable free software gilded with AI sparklies. But the costs? Staff, overhead, and whatever perks go with being a wizard at a Chinese university have to be counted, multiplied by the time required to get the system to work mostly, and then included in the statement of accounts. These steps have not been taken, but a company named Complete Analytics should do the work.
Second, what was the cost of the social media campaign that made Deepseek more visible than the head referee of the Kansas City Chiefs and Philadelphia Eagle game? That cost has not been considered. Someone should grind through the posts, count the authors or their handles, and produce an estimate. As far as I know, there is no information about who is a paid promoter of Deepseek.
Third, how much did the electricity to get DeepSeek to do its tricks? We must not forget the power at the universities, the research labs, and the laptops. Technology Review has some thoughts along this power line.
Finally, what’s the cost of the overhead. I am thinking about the planning time, the lunches, the meetings, and the back and forth needed to get Deepseek on track to coincide with the new president’s push to make China not so great again? We have nothing. We need a firm called SpeculativeAnalytics for this task or maybe MasterCard can lend a hand?
Net net: The Deepseek operation worked. The recriminations, the allegations, and the explanations will begin. I am not sure they will have as much impact as this China smart, US dumb strategy. Plus, that SemiAnalytics’ name is a hoot.
Stephen E Arnold, February 7, 2025
Online Generates Fans and Only Fans
February 6, 2025
Ah, the World Wide Web—virtual land of opportunity! For example, as Canada’s CBC reports, "Olympians Are Turning to OnlyFans to Fund Dreams as they Face a ‘Broken’ Finance System." Because paying athletes to compete tarnishes the Olympic ideal, obviously. Never mind the big bucks raked in by the Olympic Committee. It’s the principle of the thing. We learn:
"Dire financial straits are leading droves of Olympic athletes to sell images of their bodies to subscribers on OnlyFans — known for sexually explicit content — to sustain their dreams of gold at the Games. As they struggle to make ends meet, a spotlight is being cast on an Olympics funding system that watchdog groups condemn as ‘broken,’ claiming most athletes ‘can barely pay their rent.’ The Olympics, the world’s biggest sporting stage, bring in billions of dollars in TV rights, ticket sales and sponsorship, but most athletes must fend for themselves financially."
But wait, what about those Olympians like Michael Phelps and Simone Biles who make millions? Success stories like theirs are few. The article shares anecdotes of athletes who have taken the Only Fans route. They are now able to pay their bills, including thousands of dollars in expenses like coaching, physical therapy, and equipment. However, in doing so they face social stigma. None are doing this because they want to, opines Mexican diver Diego Balleza Isaias, but because they have to.
Why are the world’s top athletes selling (images of) their finely honed bodies to pay the bills? The write-up cites comments from the director of Global Athlete, an athlete-founded organization addressing the power imbalance in sports:
"’The entire funding model for Olympic sport is broken. The IOC generates now over $1.7 billion US per year and they refuse to pay athletes who attend the Olympics,’ said Rob Koehler, Global Athlete’s director general. He criticized the IOC for forcing athletes to sign away their image rights. ‘The majority of athletes can barely pay their rent, yet the IOC, national Olympic committees and national federations that oversee the sport have employees making over six figures. They all are making money off the backs of athletes."
Will this trend prompt the Olympic Committee to change its ways? Or will it just make a rule against the practice and try to sweep this whole chapter under the mat? The corroding Olympic medals complement this story too.
Cynthia Murrell, February 6, 2025
-
- Subscribe to Beyond Search
Feature archive
News archive
-
