Google Knows It Is Winner Take All: Go Go Go
September 23, 2026
Another dinobaby post. If there is art, it AI generated. I am Grandpa Arnold, not Grandma Moses.
Here are Beyond Search, we love Google. We love the loss of Boolean. We love the specious AI intrusions. We love the deplatforming of YouTubers explaining gardening. We love the advertising (just like the US courts do but more intensely). We love the personnel policies. We love the management steadiness. We love the customer support. We love the documentation written in a language everyone should learn. It is called “Googlese.”
Thanks, Midjourney. Good enough.
Most of all we love Google’s love of explaining the obvious greatness of Google; to wit: “Google Just Dropped a Bombshell on AI Spending. Alphabet’s $200 billion AI Bet Is Starting to Pay Off.” Barchart delivers this Googley information without the encomia I think the Google deserves.
The write up says:
Google Cloud CEO Thomas Kurian just supplied an unusually concrete answer. Speaking at the Goldman Sachs Communacopia + Technology Conference, Kurian said the aggregate payback period on Google’s AI servers is less than two years. More importantly, he said the payback period for Google’s own silicon is roughly half that — implying a period of less than one year.
Do Googlers wear T shirts with the word “Winner!” on them? Not only has DeepMind Google broken into self learning smart software, the company continues to innovate in saloon door online advertising mechanisms. Google is just doing outstanding work. Plus, legal decisions in the US continue to do the Googley stuff. Google does not have to dismantle its online advertising machine. Few pay attention to the payment rails starting to form from the China-certified Alibaba AliPay system. The only downside to Google taking over the world is the pesky European Union dullards.
What is Google’s secret to Cloud domination? According to the write up:
Google does not have to buy every AI accelerator from Nvidia (NVDA). It has spent years developing its own Tensor Processing Units (TPUs), allowing it to design chips around its specific workloads. Kurian said Google’s accelerator business, including TPUs, is already more than twice the size of the next-largest hyperscaler’s accelerator business. He also said Google offers 2.7 times better price performance for AI training and 80% better price performance for inference, compared with the relevant alternatives. That helps explain why Google’s enormous capital budget does not necessarily represent an enormous financial burden, despite producing negative free cash flow last quarter for the first time ever. If a server pays for itself in less than two years — and proprietary silicon pays back in roughly one year — Google can recycle that capital into the next generation of infrastructure much faster than a simple comparison of annual capex suggests.
The Google is making smart software payoff because it has chips, online advertising (eyeballs and advertisers), Baskin Robbins-type cloud services, and a line up of units, subsidiaries, partners like those fun loving folks at Alibaba, and much more. Google is vertically integrated just like Henry Ford’s River Rouge operation for auto assembly was. Coal and iron ore come in and nifty cheap autos roll out the other end of the “factory.”
I used to work at what once was a blue-chip outfit engaged in general business and technology consulting. In the day-to-day interactions with people who were described to me as the smartest in the world, I learned that the River Rouge model was important. It sold cars, but it was a bit of a problem for the ducks and fish near the facilities’ discharge pipes that dumped waste into the confluence of the Rouge River and the Detroit river. Once operational, the River Rouge facility, according to some of the world’s smartest MBAs was inflexible. Change at some levels with easy; at others, impossible. With most functions in one vertically-integrated stack, even Henry had to figure out how to maintain and manage the operational complexity of the manufacturing stack. The investments in infrastructure were hefty, and the Ford outfit had to make sure that money kept flowing in. But what made the original Model T the cat’s pajamas required some strong engineering capabilities and managerial (maybe somewhat crazy mono-maniacal management methods) became more tricky in the sprawling River Rouge facility. That worked until Alfred P. Sloan set up a decentralized, flexible manufacturing model. Ford found itself Number 2. Sure, it made money but vertical integration is somewhat less nimble that Sloan’s set up. Since I majored in medieval poetry, I just accepted what those slick MBAs spouted at me at the blue-chip firm.
Google’s approach seems to be Henry Ford, River Rouge-play. Google doesn’t like to lose in advertising, in AI, or court. The cited article concludes:
Alphabet’s roughly $200 billion capex plan for 2026 looks far less frightening if Google’s AI infrastructure really pays back in under two years — and its own silicon in roughly one year. Granted, Kurian’s figures are company disclosures, not guarantees. AI demand could slow, chip economics could deteriorate, or competitors could close Google’s performance gap. Yet, the latest numbers point in the opposite direction. Google Cloud revenue grew 82% in Q2, backlog reached $514 billion, and Google’s custom accelerator business is already more than twice the size of the next-largest hyperscaler’s accelerator business. GOOGL stock trades at roughly 16 times both trailing and forward earnings, with a PEG ratio of less than 1 times. Wall Street maintains a consensus “Strong Buy” rating for GOOGL stock based on 54 analysts with coverage. For investors, that makes Alphabet look less like a company simply spending $200 billion on the AI boom and more like one building an AI infrastructure machine designed to pay for itself. Alphabet remains one of the more compelling ways to participate in AI spending without betting entirely on someone else’s chips.
Can Google avoid the types of issues that allowed Mr. Sloan to become number one? Of course. Google is stuffed with the smartest people in the world. Google is fast. Google is flexible. Google is diversified. The decentralized and distributed approach is already integrated into the Googley machine.
What could go wrong? Here at Beyond Search we believe that Google is not subject to the lessons of the past learned by studying Henry Ford’s obvious failure and demonstration of ill-advised management decision making. Loser! But Google? Winner!
Stephen E Arnold, September 23, 2026
Does a Data Center in a War Zone Need a Red Target on Its Roof? Nah, the Huge White Roof Shouts, “Target”
September 21, 2026
Another dinobaby post. If there is an image, AI did it. I am Grandpa Arnold, not Grandma Moses.
The Wall Street Journal reports “AWS Says It Can’t Restore Some Data from Mideast Facilities Struck by Iran.” Imagine that. I hopped online and snagged a generic picture of a data center viewed from the air.
Source: Russian Alibaba, September 17, 2026
What are the characteristics of this generic data center? Here’s my Monday morning quarterback observations. It looks big. It seems to be a shoebox type of design. It has a white room. It has outbuildings for odds and ends.
How hard would it be for a teenager in the army of a power hostile to the US to fly a drone with a kinetic to the roof of this structure and set off the kinetic charge? The answer is, “Pretty easy.” What if there were so-so missiles with close enough for horse shoes guidance systems? Several missiles could do the job.
I have driven by some Fancy Dan data centers; for example, the facility in Ashburn, Virginia, where one of my US government projects ran. The front and sides of the structure were hardened. But the roof? Nah. Why not bury the data centers following the Iron Mountain approach to document storage? Cost. Why not harden the roof? Cost. Why not equip data centers with anti-drone and anti-missile defenses? Cost and maybe some pushback from neighbors with no interest in having missile launchers near their subdivision.
So where are we? Here are a few comments from the YCombinator Hacker News’ comments about the Wall Street Journal story:
- cmiles8 posted: This interview with an AWS leader isn’t aging well, from CBS Sunday morning: Pogue asked, “I don’t mean to give anyone ideas, but let’s say I figured out that one of these unmarked buildings was an AWS data center, and I blew it up. Are you saying that it’s so backed up and redundant that you probably wouldn’t notice?” Wood replied, “Yeah, you wouldn’t notice. I mean, we might be a bit upset, but you wouldn’t notice!” https://www.cbsnews.com/news/cloud-computing-loudoun-county-…
- avereveard posted: Aws cto is basically opening every presentation with a everything breaks all the time slide, idk where you get that idea they oversell the cloud resilience
- jaggerest posted: I feel like if this was done correctly it would eat a bunch of the market, but I question how many people are actually willing to pay for “the right way”.
- cyberax posted: AWS is designed to survive one datacenter being offline (which happened more than once, btw). When the first DC in ME was hit, AWS continued working normally, with only a few services experiencing issues. But it’s not designed to survive TWO datacenters going offline, and in a permanent fashion.
- wmf posted: They [Amazon] did build redundancy but most of it was bombed.
- betelbuddy posted: AWS always made very clear they wont copy your data to another region as only you know what your compliance and data residency requirements are. But at the same time they always said, its up to you to come your with your disaster recovery strategy based on your project requirements. And it has always been the case copying your critical data to another region is one of the first things on your check list.
- gilbetron posted: Force Majeure. Except for payment obligations, neither party nor any of their affiliates will be liable for any delay or failure to perform any obligation under this Agreement where the delay or failure results from any cause beyond its reasonable control, including acts of God, labor disputes or other industrial disturbances, electrical or power outages, utilities or other telecommunications failures, earthquake, storms or other elements of nature, blockages, embargoes, riots, acts or orders of government, acts of terrorism, or war.
Net net: The killer comment is the CBS Sunday morning item. I seems the statement “You wouldn’t notice” is not entirely accurate.
Stephen E Arnold, September 21, 2026
Coincidence: Unlikely
September 7, 2026
On September 3, 2026, something interesting seems to have taken place. OpenAI, Anthropic, xAI, and Google learned that their AI systems experience a digital heart failure. The hearts began beating again, but the leadership of these firms had a headache, and I am not sure it has gone away.
Ars Technica said in “Four Major AI Models Suffer Rare Overlapping Downtime. Service Interruptions Hit ChatGPT, Claude, Grok, and Gemini Practically Simultaneously.” How about this for a velvet glove observation?
While the affected frontier models go down occasionally, having all four experience interruptions in the same short period is practically unheard of. Claude reports 99.4 percent uptime for its services over the last 90 days and last reported a similar three-hour “partial outage” on August 24. OpenAI reports 99.63 percent uptime for ChatGPT and 100 percent uptime for ChatGPT Codex in the same period. ChatGPT’s so-called “Work Mode” reported an hours-long period of “elevated latency” on August 31.
Computerworld in “ChatGPT, Claude, and Grok All Went Down at Once; Enterprises Need a Backup Plan” stated:
…on Thursday, as OpenAI’s ChatGPT, Anthropic’s Claude, and SpaceXAI’s Grok near-simultaneously, and somewhat mysteriously, experienced significant, prolonged outages.
Yep, mysterious.
Let’s stop and ask a handful of questions not addressed in these two cited write ups:
- What happened to the cloud providers and their systems? Don’t infrastructure operators have smart software tuned to identify and fail over immediately? Everyone’s favorite BAIT (big AI tech company) outfit not only owns its vertical stack, the firm owns a couple of state-of-the-art cyber security systems. What went wrong in the engineering, design, and implementation of these smart defense mechanisms? I think I know the answer. A lawyer will help craft a plausible deniability statement or tell the clients with the problematic system, “Keep you lip zipped.”
- What about the tech bloggers and YouTube experts? Where are those deep dives into the technical issues that caused for BAITs to go south at approximately the same time? (Is it possible that analyses have been posted and remain unindexed in order to convert the story into the information equivalent of a squirrel dashing into a roadway only to be squished by a Ford F-150’s front tire?)
- What are the BAIT ourfits saying? Yeah, not much, right? Does the cat have your PR departments tongue?
I haven’t been in a statistics class in more than half a century. The idea that four unrelated companies’ smart software glitches at about the same time on the same day seems unlikely. Is there an undisclosed or previously unknown dependency sparking this dumpster fire? Did an outside actor send a signal to test the efficacy of its “kill US AI services” switch? I can think of a couple of countries who might have explored developing such a software device.
I used the free pen I snagged at the Staunton, Virginia, Comfort24 Hotel. I roughed out some numbers. I recast these into one of my analogies. For four unrelated AI firms to experience an outage at roughly the same time on the same day of the week is like a public park with four kid-oriented merry go rounds rotating 24 hours a day. Each merry go rounds dumps kids off. The chance that I could look at the four rotating merry go rounds and see each of the four dumping a kid simultaneously is about one in 6,000,000. For me, the coincidence is sufficiently improbable to say, “Yo, we have a shared dependency or we have a bad actor testing a kill switch.”
In my typical dinobaby fashion, I wish to offer a few observations. [a] Sheer chance is unlikely to produce this type of event. [b] A bad actor working from a position with tools that permits a kill actio to occur seems worth considering. [c] The baloney about security and reliability needs a footnote to two to allay the concerns of suspicious individuals like me.
Net net: We just received a signal. Who will stand up and say, “Okay, folks this is what really happened.” Tip: Don’t hold your breath.
Stephen E Arnold, September 7, 2026
The Cloud Is Great Until the Tech Bros Go South and Use Legal Eagles Instead of Common Sense
August 17, 2026
Another dinobaby post. No AI unless it is an image. This dinobaby is not Grandma Moses, just Grandpa Arnold.
I found this news item in Current.org a signal about a horizon event. Okay, what’s a “signal.” In my lingo this is an observable or capturable event that when grouped with similar items can be a predictor of the future. What’s a “horizon event.” In my lectures, a “horizon event” is an action that seems (based on available information) to suggest that a certain outcome is likely to occur in the next six to 18 months. You can critique my approach, but it works for me. Don’t like my method? Hire someone who lectures at the Institute of the Mathematical Sciences of the America.
Thanks, MidJourney. Good enough.
The article is “Nine PBS Sues Iron Mountain Over Blocked Access to Archival Data”. The essay explains that a public TV / radio outfit cannot download its content. The organization’s online provider went out of business. That online provider is one of those enablers that hook a broadcast outfit to a third party running a system on which the broadcaster’s data reside. The intermediary went south (that is, failed as a business) and the third party provider says, “Hey, your public TV crowd cannot access your data.”
Thanks, MidJourney. Good enough.
This is 21st century code for letting lawyers resolve the problem. Common sense suggests that the third party vendor (an outfit that once made money putting paper in caves) locked the people with nice hair out of their data. The write up says:
Iron Mountain has refused to return the materials to the station because its client, OSS, technically owned “the physical services housing the data” within Iron Mountain, according to the complaint.
Nice thinking by an outfit comfortable putting boxes filled with documents in cold storage. How long has Iron Mountain had this broadcaster as a client? Just decades. Plus those public broadcast outfits are tricky with their membership drives, public service announcements, and distinctive speech styles. Definitely up to no good so Iron Mountain has to be really, really careful.
And Iron Mountain knows it is just doing what its lawyers suggest:
Iron Mountain did not respond to a request for comment.
So what’s the signal? I think there are several in this anecdote:
- A cloud vendor may be a chain of companies. The client deals with one link and is essentially clueless about what vendor possesses the client’s data. Rules of access are not discussed so the broadcaster is at fault. Common sense suggests otherwise, but this is 2026, and you know what that means.
- Putting data in once place is a very bad idea. I have an aversion to the cloud because I spent years wrangling with Dialog Information Services about the data my employer uploaded to the third party system. Nope. Exclusives are handing control and power to the vendor. That’s not going to happen. We placed the same data on multiple online services. If one used Iron Mountain logic, no big deal.
- Who will own any data input to an online service that features smart software or interactions with large language models and a client’s agents or humanoids? Answer: Unless the user adds explicit language to the contract, the vendor will control access. Examples of chat inputs are reported to be turning up in the darnedest places like Web search results. You don’t own anything unless you have a legal document that says you do.
Net net: Cloud services may seem cheap, benign, caring, and friendly today. On the horizon those cheerful entities become Darth Vader’s nasty brothers, sisters, and cousins. Cloud vendors have light sabers. Cloud customers have legal filings.
Stephen E Arnold, August x, 2026
Google Exposes a Paw and a Vulnerability
May 20, 2026
Another dinobaby post. No AI unless it is an image. This dinobaby is not Grandma Moses, just Grandpa Arnold.
Google is a BAIT outfit. “BAIT” means in my lingo “big AI technology.” The companies upon which I bestow the moniker are different from the run-of-the-mill Silicon Valley outfit for several reasons:
- The stakes are very high. One company will eventually become the sun in the AI universe; other BAIT outfits will orbit it.
- Innovation is becoming increasingly competitive. The reason is that very few BAIT outfits have done much more than wrapped software around Google’s “Attention Is All You Need” essay and the fact that most of the people in the world use Google products and services. That’s gravitational pull.
- Traditional management methods for a BAIT outfit just don’t work very well. Whether it is the bonkers trial between the luminaries of Grok and OpenAI or the somewhat wonky effort to build data centers when zoning, power, and water are not cooperating, BAIT outfits are stirring up chaotic storms.
- The total win mentality throws out the rule book for everything that a traditional business once cherished.
Two entrepreneurs get a cloud invoice. These innovators can’t pay for food. A giant bill for cloud services means their dream has done up in smoke. Thanks, Venice.ai. Good enough.
I don’t want to write about Google’s human resource challenges. Personally I don’t think the AI protest roiling Google’s UK unit is going to end with the Prince falling in love with Snow White. I am not sure that Google’s injection of more and more advertising is going to keep customers happy or eagerly paying money to watch cable TV Google style. I am also skeptical that Google’s method of shoving AI into its nooks and crannies will deliver user happiness. I had to terminate a simple test of Gemini on May 11, 2026, because Gemini kept crashing and losing data. Yep, nice work, Gemini.
Instead I want to point to an article in Cybernews titled “Google Cloud Developers Going Bankrupt Over Gemini API Key Abuse: Hard Spending Caps Now Available.” The article asserts:
The Google Cloud’s subreddit has turned into a bottomless pit of people wailing over massive cost overruns and spending caps that don’t cap spending. And the cited sums are devastating. Here are just a few of the headlines posted over the past months:
- Went to bed with a $10 budget alert. Woke up to $25,672.86 in debt to Google Cloud.
- 80,000 NOK ($7,500) drained from my Google Cloud account in 5 minutes – full forensic breakdown of how the attack worked.
- Charged $10,138 in March 2026 due to Google’s documented Gemini API key vulnerability – support closed my case twice, saying “no fraud found.”
- WARNING: Google Cloud/Gemini API ”Spend Caps” do NOT work in real-time ($1,800 charged on a $100 cap).
- Google Cloud detected $975 of API key fraud on my account, sent one email at 11 p.m., then let the bill grow to $18,596 – 5 support agents have refused to help.
- $10 budget alert – hijacked Gemini API Key billed $1,300 in a few minutes.
I recall similar tales of woe from the early days of the Internet. People set up sites and then opened their monthly invoice to discover the fascinating pricing mechanisms based on tiers. As traffic went up, so did the online fees. I also recall tales of woe related to Amazon’s cloud services. Due to the brilliant interface and outstanding documentation, an AWS customer would receive a “surprise”; for example, a low cost service explodes to a high cost service without warning. The trick is that some AWS users had no idea what happened. Well, welcome to the exciting world of unilateral billing policy changes.
After doing a couple of projects for what was the old AT&T and Bell Labs outfits, I learned that these billing mechanisms were neither new or out-of-bounds. What happened is that more people than ever jumped into online and learned about pricing thresholds, special services, and tiering really meant: Money for good old Ma Bell. Yes, there was a reason the old AT&T morphed into the wonderful oligopolies we enjoy today in the US of A.
Now back to the write up in Cybernews. I noted this passage:
It appears Google Cloud has no hard spending caps, and its fraud detection tools, while capable of flagging suspicious activity, do not take automated actions to stop the abuse.
Cybernews continued:
In February [2026], Truffle Security discovered that old Google API keys, previously used in other projects as harmless identifiers, overnight became ticking time bombs once they were granted access to the Gemini API. Thousands of multipurpose Google API keys can be found exposed on websites, code repositories, apps, and elsewhere, and Google itself previously encouraged users to “safely embed them in client code.” Truffle Security even demonstrated the attack by using Google’s own exposed API keys to hit the Gemini API, and found thousands of API keys belonging to major financial institutions and other companies. “If the vendor’s own engineering teams can’t avoid this trap, expecting every developer to navigate it correctly is unrealistic.”
Google pushes actions down. People move to other teams. The result is, “Yo, no kidding.” Without a functional customer support system and spotty or non-existent documentation, using Google can be an interesting problem for a customer. When a surprise invoice arrives, the institutional memory of who, what, why, when, and how seems diaphanous or a chimera.
What’s the fix? My interpretation of the write up, is, “The customer is responsible.” Yep, global penetration to most countries on earth with billions of users. “Hey, user, it is your job.” The approach is convenient, and it explains in part the thrills and chills of relying on Google.
As I said, Gemini did not work on May 11, 2026. It is my fault. Pay your bill. Oh, and the AT&T pricing model. Former Bell and AT&T employees once worked at the Google. Their contributions are numerous just like the digits on some of those surprise invoices.
Stephen E Arnold, May 20, 2026
France Lines Up to Gore Some US Tech Bros
April 13, 2026
Another dinobaby post. No AI unless it is an image. This dinobaby is not Grandma Moses, just Grandpa Arnold.
France has élan. My interest in the reactions of people who visit the country for the first time is keen. We live in rural Kentucky, and at get togethers which often involve cooking a freshly killed pig or wearing Derby pins and drinking local beverages, tales of travelers’ experiences are exchanged. Some people love the food, but report the waiters ignored them. Other say, “I entered a bakery and said, ‘Hello,’ and no one waited on me.” The tales of getting a hefty fine on the metro, visiting a small town and finding the shops closed for lunch, and the bafflement when a Kentuckians’ passport and wallet are stolen a few minutes after landing at the Marseille France Airport are far too common.
Venice.ai, you don’t know a French gendarme from a guardrail. Good enough.
Imagine how Americans working at Microslop, sorry, I meant to type Microsoft are reacting to this French decision: “France Says “Au Revoir” to Windows, “Bonjour” to Linux.” To be fair, I don’t know if French people who use Microsoft products will switch. The government agencies will adapt over time. Nevertheless, the decision is delightfully French.
The cited article reports:
France is planning a major shift in its government technology infrastructure, announcing its intention to move away from Microsoft Windows in favor of Linux. The decision marks a significant step in the country’s broader effort to reduce reliance on U.S.-based technology companies and regain control over its digital systems. The transition will begin with government workstations, particularly within key digital agencies, as part of a wider strategy to adopt open-source and locally controlled technologies.
From my point of view, this is just one example of the steps France and other countries will be taking to wean themselves from the tech bros’ approach to computers, online information, and software. Pressure is likely to be exerted on French government contractors to ensure that “digital sovereignty” is part of the work process.
What’s this mean for French technology firms? I would assume that big outfits like Dassault Systèmes-type organizations will comply, probably on a negotiated timeline. Private startups will use whatever is cheaper and does the job. If that means, Microsoft PowerPoint, those firms will do what they can until France ratchets up the pressure. Who can forget the arrest of Pavel Durov in August 2024. That action appears to have put a hitch in the Telegram git-along.
The write up points out:
Government leaders have emphasized the need to regain control over national data, infrastructure, and decision-making systems, rather than depending on foreign technology providers.
The write up adds:
France’s move highlights a growing shift in how nations view technology – not just as a tool for productivity, but as a strategic asset tied to sovereignty, security, and long-term independence.
My take: More bad news for US technology companies is coming. Next up will be actions directed at US firms’ business practices. In fact, at some point, I can see an American tech bro landing his private jet at a French airport. He will be detained and charges will be filed. He will be able to leave France, but the French wheels of justice will grind forward. Many things can result from this type of French direct action.
Today Linux Tomorrow maybe a Silicon Valley luminary. Just a thought.
Stephen E Arnold, April 13, 2026
A Pragmatic Look at Data Center Power: Just Turn Everything on at Once. No Problemo.
March 12, 2026
Another dinobaby post. No AI unless it is an image. This dinobaby is not Grandma Moses, just Grandpa Arnold.
A happy honk to the UK online information service the Register. “Your Datacenter’s Power Architecture Called. It’s Not Happy” presents some useful information about how AI compute data centers operate when the “on” button is pressed. The article explores in reasonably a non-EE geeky way some of the different facets of taking a corn field in Louisiana or a tobacco farm in Tennessee and building a five soccer field size data center. Heck, make it bigger. Land is cheaper than in Manhattan. Go for 20 soccer fields of AI compute hardware.
Yep, AI is close enough for horseshoed. Thanks, MidJourney. Nice fire.
The write up says:
GPU clusters and AI accelerators don’t operate on the old rules. They don’t ask for 15 kW. They demand hundreds of kilowatts per rack, an order-of-magnitude leap that legacy electrical and thermal architectures were never designed to survive. The comfortable assumptions baked into decades of datacenter design are now liabilities, and the industry is facing a reckoning it can no longer defer.
What? Are big time AI outfits kicking the can down the road or simply ignoring the fact that “power” is just there. It’s like water. In their accelerated lives, power has not been top of mind. Now the whiz kids at big tech AI companies have an opportunity to learn about the challenges their acres of computers and assorted gizmos will require.
The write up identifies a number of issues; for example:
Let’s talk about the current-squared problem and resistive losses. Because power loss scales with the square of the current, even small reductions in current lead to significant increases in efficiency. The power distribution efficiency is governed by Joule resistive loss (Ploss</loss> = I2R).
If you are not into this power lingo, the author seems to be saying that the new AI compute data centers require some extra engineering. Why? The pricey CPUs and GPUs can run hot. These devices plus RAM and solid state storage are voltage?sensitive. Even idle or under utilized racks can run at higher temperatures than old-fashioned data center racks. When demand hits to create a video for grandma, the AI compute system has to deal with heat. AI centric systems can throttle themselves. Pushing harder means that electrical noise can become a factor. Furthermore, failover is harder because the stacks are non?identical and latency?sensitive.
How do you deal with infrastructure challenges, power, and cooling? Answer: Novel engineering and money. The problem is that these big AI compute data centers are now the equivalent of the room-sized mainframes in the late 1950s. The solutions make 20 somethings laugh when the Smithsonian in DC puts some of its historical hardware on display.
The real equations are not the EE marvels. Nope, the basic equations pivot on the cost of time required to engineer, design, test, and manufacture the components necessary to keep power draw lower and temperatures even lower. At the same time, the solutions have to allow the CPUs and GPUs to go fast.
Accelerationists think about software. Accountants and people with common sense think about plumbing. Experienced professionals think about time and how much it costs to crunch quite challenging engineering into tiny intervals. Speed can kill data centers, the financial dreams of high tech whiz kids, and some businesses.
What happens when one tries to scale these nifty data centers? No big deal. We are big tech doing AI. We have the answers.
Stephen E Arnold, March 12, 2026
Google Finally Has a Winning Play in the eGame Sector
February 5, 2026
Another dinobaby post. No AI unless it is an image. This dinobaby is not Grandma Moses, just Grandpa Arnold.
Google has had interest in online games, gaming, eGames, and related pastimes for a couple of decades. But control of online games like the online ad agency’s control of Web search has eluded the friendly outfit. I recall Google’s interest in an online gaming tie up in the early 2000s. I don’t think I heard much about that other than it was a non-starter.
You may or may not recall such Google efforts as Android gets the default Google Play store as a way to “get” games. Google did a service within YouTube focused on eGames. Google fiddled around with a social media play in eGames which was part of the Google+ service. There was, I believe, there was a me-too of Apple’s Game Center for Android. About a decade ago, the Google tried to buy Twitch, showing that gamer eyeballs seemed important in 2014. In 2019, Google rolled out the barrels of money and created Stadia. The store and other parts of the service were just killed. Someone at Google rounded up “leadership” support for a Google Play Pass, and I still don’t know what that provides. There is something called “Game Snacks.” Plus the Stadia “technology” is floating around in the Google Cloud or was the last time I checked. I think what I remember is that Google was doing its Don Quixote approach: Keep going until one kills a windmill… in an eGame or a Google wizard’s mind.
A large creature named Googzilla is really excited about its stumbling into a greenfield with bundles of cash to pick up and use. There’s no competition in sight. Now all the creature has to do is wait for the developers, eGame companies, and players to come to this field and have fun. The creature will just gather up the money, of course. Thanks, Venice.ai. Good enough.
Google may have made its mark on gaming.
I read “Videogame Stocks Slide on Google’s AI Model That Turns Prompts into Playable Worlds.” The trusted news source’s story says:
The AI model, dubbed “Project Genie”, allows users to simulate a real-world environment through prompts with text or uploaded images, potentially disrupting how video games have been made for over a decade and forcing developers to adapt to the fast-moving technology…. Project Genie also has the potential to shorten lengthy development cycles and reduce costs, as some premium titles take around five to seven years and hundreds of millions of dollars to create. Videogame developers have been increasingly adopting artificial intelligence as a way to stand out in a highly competitive industry dominated by large players. A Google study last year showed that nearly 90% of game developers use AI agents.
I interpret this to mean:
- Google has taken a step toward becoming the go-to source for some of the most expensive and tedious part of game development
- Google has made it clear that as its AI improves, the smart game outfits will use Google services
- Consumers of eGames will look for game experiences enhanced, built upon, and generated by Google. (Why not just do everything related to the eGame in Google?)
Adding up these points, Google may have found a way to begin to expand its eGame capabilities attracting independent developers, forcing eGame giants to realize the value of their firms and products can be affected by the long-time eGame fumbler, and disrupting the how of eGames.
Net net: This is a big deal. My hunch is that Google finally rolled out something that would allow the firm to increase its eGame revenue. In my opinion, Googzilla stumbled into a green field of money. How long will it take regulators and consumer advocates to understand the impact of this new eGame success? A long, long time.
Stephen E Arnold, February 5, 2026
Microsoft: Parallel Wagering and Risk
January 30, 2026
Another dinobaby post. No AI unless it is an image. This dinobaby is not Grandma Moses, just Grandpa Arnold.
I had a French teacher who loved French proverbs. Her name was Madame Matthews, and she was a German teaching French in a one-horse town in central Illinois. I vaguely recall that one of her lessons was a knock off of what I thought was a British proverb. Here it is in la langue de Molière:
Le jeu est le fils de l’avarice et le père du désespoir.
However, I just read “Microsoft Lost $357 Billion in Mark Cap As Stock Plunged Most Since 2020.” That’s a big bet Satya Nadella made, and it has been working. Until now. Then I read “Microsoft Shares Dive as Data Center Spending Overshadows Earnings Surge.” Yep:
Le jeu est le fils de l’avarice et le père du désespoir.
Big bets. Big risk.
A wealthy professional is gambling at multiple games of chance at the same time. A crowd is watching, amazed that this person would play blackjack, run the slots, and hope that the roulette ball in on his wave length. Thanks, Venice. Good enough.
Now let’s put this swing for the fences strategy in the context of a run of the mill user. “Microsoft Is Working to Rebuild Trust in Windows” provides some consulting type lingo to the “experience” of using the world’s most popular operating system. That article offers:
Windows is at breaking point, and Microsoft knows it. Sources familiar with the company’s plans tell me Windows engineers are now focusing on fixing the core issues of Windows 11 over the coming months, in a process known as “swarming.” Microsoft is redirecting engineers to urgently fix Windows 11’s performance and reliability issues, aiming to halt the operating system’s death by a thousand cuts.
Let’s think about Microsoft and the interesting idea of “fixing the core issues.”
Microsoft is the juicy target for many bad actors. Some of the vulnerabilities have been known and exploited for years. In Tallinn, Estonia, I spoke with an advanced computer science class. One of the students said, “Yeah, Windows is pretty easy to exploit.” When was I in Estonia? How about 2006? The class was learning what flaws Windows had because it was a big, fat, floppy target.
Next let’s consider that Microsoft is using AI to code its different services. The only problem is that some of that code output by smart software has flaws. Humans may not be available to grind through the outputs searching for snipes. My recollection is that snipes were quite difficult to locate.
Also, consider that Microsoft has allowed interns, contract workers, real live on site programmers to add features and functions to Windows that demonstrate progress, commits, or innovations for the sole purpose of getting a raise, bonus, or promotion. I gave a lecture at Microsoft after which there was a reception. I spoke with a senior lead something in the Word unit. I asked if the word processor would ever include the type of numbering system used by the law firms with which I interacted. He said, “No. No need.” Right! No need. There were specialists like the little outfit in Charlottesville, Virginia, which figured out how to make Word number the way lawyers and courts want documents numbered. That little outfit made a lot of money selling its widget to law firms. Microsoft was right, “No need.”
And to end this list of issue, allow me to point out that the interface across the heavy lifting apps like PowerPoint and Visio is not normalized. Exactly how does one add specific tools to custom toolbar without first mastering the art of solving the Riemann Hypothesis? Microsoft tries but seems to lose sight of the duck it is trying to hit with its over engineered incentive system.
Now let’s think about the French proverb. My view is that Satya Nadella has gambled, bet a lot, and is now thinking, “Now what?” Investors appear to be thinking, “Yeah, now what?” With layoffs looming, some Microsofties are thinking, “Okay, now what?”
Talk about energizing Google and the makers of Linux distributions aimed at consumer computer users is amping up. Okay, Mr. Nadella, “Now what?” But more AI? Build more data centers and carry the cost of electricity so consumers are not burdened? Do the Elon thing and power the data centers with turbines or just install one of those July 2026 mini nuclear reactors?
This proverb is one that Madame Matthews might suggest you analyze in 500 words:
Le jeu est le fils de l’avarice et le père du désespoir.
Mr. Nadella is not playing one game. He is parallel betting. Does that reduce risk or increase it? Why not ask OpenAI?
Stephen E Arnold, January 30, 2026
Amazon AWS: Two Pizza Team Engineering Delivers Indigestion to Lots of People
October 20, 2025
No smart software. Just a dumb and quite old dinobaby.
Years ago an investment bank asked me to write a report about Amazon’s technical infrastructure. I had visited Amazon as part of a US government entity. Along with four colleagues from different agencies, I had an opportunity to ask about how Amazon’s infrastructure could be used as an online services platform. I did not get an answer, just marketing talk. One of the phrases stuck with me; to wit, “We use two pizza teams.”
The idea is that no technical project can involve more developers than two pizzas can feed. I was not sure if this was brilliant, smart assery, or an admission that Amazon was a “good enough” engineering organization.
I had a couple of other Amazon projects after that big tech study. One was to analyze Amazon’s patents for blockchain. Let me tell you. Those Amazon engineers were into cross chain methods and a number of dizzying engineering innovations. Where did that blockchain stuff go? To tell the truth, I don’t have many Amazon blockchain items lighting up my radar. Then I did a report for a law enforcement group interested in Amazon’s drone demonstration in Australia. The idea was that Amazon’s drone had image recognition. The demo showed the drone spotting a shark heading toward swimmers. The alert was sounded and the shark had to go find another lunch spot. What happened to that? I have no idea. Then … oh, well, you get the idea.
Amazon does technology which seems to be okay with leasing Kindle books and allowing third party resellers to push polo shirts. The Ring thing, the Alexa gizmo, and other Amazon initiatives like its mobile phone were not hitting home runs.
I read “Widespread Internet Outage Reported As Amazon Web Services Works on Issue.” [This is a Microsoft link. If it goes dead, don’t call me. Give Copilot a whirl.] Okay, order those pizzas. The write up reports:
The Amazon cloud computing company, which supports wide swaths of the publicly available internet, issued an update Monday just after 3 p.m. ET saying that the company continues to “observe recovery across all AWS services.” “We are in the process of validating a fix,” AWS added, referring to a specific problem set off by the connectivity issue announced shortly after 3 a.m. Eastern Time.
Okay, that’s 12 hours and counting.
I want to point out that the two-pizza approach to engineering is cute. The reality is that AWS is vulnerable. The outage may be a result of engineering flubs. You are familiar with those. The company says, “An intern entered a invalid command.” The outage may be a result of Amazon’s giant and almost unmanageable archipelago of servers, services, software, and systems was hacked by a bad actor. Maybe it was one of those 1,000 bad actors who took out Microsoft a couple of years ago? Maybe it was a customer who grew frustrated with inexplicable fees and charges? Maybe it was a problem caused by an upstream or downstream vendor? One thing is sure: It will take more than a two pizza team to remediate and prevent the failure from happening again.
In that first report for the California money guys, I made one point: The AWS system will fail and no one will know exactly what went wrong.
Two pizza engineering is a Groucho Marx type of quip. Now we know what one gets: Digital food poisoning.
Stephen E Arnold, October 20, 2025 at 530 pm US Eastern
-
- Subscribe to Beyond Search
Feature archive
News archive
-
