đ§ Neural Dispatch: OpenAI needs a favour, Mustafa Suleymanâs superintelligence vision, and Amazonâs tiff with Perplexity
The biggest AI developments, decoded. November 12, 2025.
Hello!
Cognitive warmup. You know they know the gameâs up, but the arrogance is overpowering. Marc Andreessen, the co-founder of the venture capital firm Andreessen Horowitz, thought it was prudent to mock (in a retweet, with Pope Leo XIV, the head of the Catholic Church, after he called on AI makers to âcultivate moral discernmentâ as a fundamental part of their work. Of course, since it is Andreessen, there is an (un)healthy Sidney Sweeney obsession (though he also seems to have the EUâs Thierry Breton living rent free in his head). It is good to see Pope Leo XIV directly address the AI bros and tell them straight. Which of course the Silicon Valley folks, mostly CEOs and VCs, often donât like. Not someone who can be brushed aside either, though Andreessen tried. This is just symbolic of where AI is, the bubble becoming clearer. Sam Altmanâs OpenAI pretty much requested for the governmentâs guarantees on the AI companyâs loans, the latest in that chapter of things going pear shaped rapidly. The frustrations are clearly showing. Has anyone tried telling the AI bros âThou shalt not stealâ?
ALGORITHM
This week, we have a conversation about Microsoftâs intention for building superintelligence that keeps humans at the top of the so-called food chain, Amazon being absolutely unimpressed with Perplexityâs Comet agentic browser behaving like a human on its shopping platform, and the rising cost of electricity thanks to AI.
Superintelligent AI dreams
Microsoft AI, though too early to know for sure, suggests that their vision for superintelligence is to âensure humanity remains at the top of the food chainâ. I wouldnât have given this much credence otherwise, but for the fact that Microsoft AI head Mustafa Suleyman is saying it, detailing what he calls Humanist Superintelligence or HSI. âWe want to both explore and prioritize how the most advanced forms of AI can keep humanity in control while at the same time accelerating our path towards tackling our most pressing global challenges,â he says, talking about the newly formed MAI Superintelligence Team. Two global challenges that Suleyman talks about â medical superintelligence, and optimising workflows to make them more energy efficient. Impressive vision, weâll celebrate when even a fraction of this is tangibly achieved, without putting humans at a disadvantage. For one, AI companies could start by defining âconsciousnessâ first, a key in finding balance between humans and AIâs impedance.
Electricity costs, and political victories
There is a growing sense that soaring electricity prices in the US dictated results of the recent elections in New Jersey, Virginia, and Georgia. A lot of credit for rising utility bills lies at the doorstep of AI data centers, as well as electric motoring. Enough data suggests as much. A forecast earlier in the year by International Energy Agency (IEA) indicated that the US electricity demand to grow at an average annual rate of about 2% over the period 2025-2027. The latest census data in the country illustrates that every one in three US households has reported forgoing necessities, such as food or medicine, to pay household energy bills. There is also data that New Jersey has seen one of the largest spikes, with retail rates jumping as high as 20 percent this summer. At a time when the AI bosses are only talking about trillions of dollars of investment in data centres, the citizenry thatâs already struggling with rising costs of sustenance, is clearly getting fed up with billionaires preaching the benefits of AI. Now, the political candidates who have won the elections, will be held equally accountable if they donât deliver an improvement.
A champion of AI, unimpressed with agentic AI
Amazon has sued Perplexity AI over its new Comet, the AI browser, which has the habit of autonomous browsing and an assistant tool that can supposedly do many things. Amazon alleges that Comet can log into a userâs Amazon account and attempt to place or modify orders while disguising itself as normal human traffic. In legal terms, Amazonâs lawyers say Perplexityâs trespass involves code rather than a lockpick. Perplexity would most likely call this innovation of some sort. I wouldnât know, I strictly keep my distance from AI browsers â there is very little to trust down that path. Interestingly enough, Amazon too has developed some level of agentic AI on its shopping presence â thereâs âBuy For Meâ in some markets which lets users shop within its app, as well as the Rufus AI assistant (it shows prominently on the app in India too) that can recommend items and manage carts.
THINKING
âI want to clarify my comments earlier today. OpenAl is not seeking a government backstop for our infrastructure commitments. I used the word âbackstopâ and it muddied the point. As the full clip of my answer shows, I was making the point that American strength in technology will come from building real industrial capacity which requires the private sector and government playing their part. As I said, the US government has been incredibly forward-leaning and has really understood that Al is a national strategic asset.â
â Sarah Friar, CFO, OpenAI.
There is something almost poetic about Silicon Valleyâs new-found patriotism. The same industry that once prided itself on being âborderlessâ, and had an aura of being too busy inventing the future to deal with government paperwork or regulatory hassles, is now shaking Washingtonâs gates with the subtlety of a brass band. Theyâve one tiny favour to ask â please underwrite our trillion-dollar ambitions. Not to interfere, of course. Just to catch the inevitable free-fall. They know itâs coming. They know itâs a bubble.
OpenAIâs CFO, no matter how much she may try to backstop the backstop, wants the US government to guarantee the mountain of financing they (and their creative investment partners such as Nvidia, Oracle and AMD will need) to further inflate the bubble as much as itâd go. Or in other words, to apparently build its future GPU infrastructure empire. The logic that we are loudly hearing from OpenAI and the likes of Jensen Huang (who has seemingly paused his âlets make a dealâ world tour to make the plea) goes something like this â AI is too strategically important to fail, America must lead, and therefore taxpayers should be morally prepared to pick up the cheque if the math doesnât math. And while weâre here, could the Treasury perhaps fill out the loan guarantee form in triplicate? The OpenAI and Nvidia tag-team is very visible.
The context: The AI bros seem to be channeling the tech era version of their inner Founding Father vibes. Suddenly, GPUs are the new railroads. AI clusters are national projects. Silicon is a symbol of sovereignty. The marketing copy writes itself. Thereâs even a little flag-waving flourish at the end â If we donât build it, China will. You donât want that, do you? Specific to OpenAI is the illusion Altman has tried to create â that theyâre too big to fail. If ever such a thing existed. And even more so in this case, where the company is yet to turn a profit in its existence, has been creatively circling investment money as if no one is noticing, and absolutely botched the latest models which were supposed to usher in the âPhD-level intelligenceâ era. They still stumble over basic math. Some interesting numbers â this AI company lost $13.5 billion on $4.3 billion in revenue for H1 2025.
The AI industry has been burning money at a rate that would make WeWorkâs accountants blush. The capital expenditure required for model training, inference infrastructure, and chip supply isnât merely high â itâs almost vertical. A realisation has dawned that private capital alone wonât shoulder national AI infrastructure, to the tune of billions of dollars. The new plan, probably thought of in a San Francisco coffee shop (thatâs how deals are mostly done in that part of the world) is to palm off the risk as patriotic duty. This takes me back to 2023 when Altman, at a summit said, âItâs hopeless to compete with usâ.
A Reality Check: When an industry truly believes in its future, it takes its own financial risks. In the case of AI, that is absolutely not the case, and therefore the subtle movement to palm it off to the unsuspecting public to absorb them. The request for backstops is not a declaration of confidence. AI companies, be it OpenAI, Nvidia, Oracle, AMD, Anthropic, Meta and Microsoft, are essentially writing a confession that they have no idea what theyâre doing, that whatever they are doing is costly, and letâs see where this takes us, as long as you split the bill. If this bubble holds, Silicon Valley will be hailed as the vanguard of a new technical era. If it doesnât, taxpayers will be left holding the bag. âWe triedâ, weâll be told. Heads they win, tails you pay.
I rarely ever agree with a politician (or an AI bro, as things shape up), but must say, it is hard to disagree with Florida governor Ron DeSantis, âMy solution is not to bail out big technology companies.â
Neural Dispatch is your weekly guide to the rapidly evolving landscape of artificial intelligence. Each edition delivers curated insights on breakthrough technologies, practical applications, and strategic implications shaping our digital future.
Excellent analysis! Your insights on the AI bubble are spot on. It's becaming evident that robust ethical frameworks, not just market forces, must guide its developement, especially for superintelligence.