Welcome to Eye on AI, with AI reporter Sharon Goldman. On this version…Meta’s large Hyperion information heart will get even greater…OpenAI and Amazon discuss an alliance…Is the AI bubble popping as a result of AI is definitely working?…AI makes inroads on the Tremendous Bowl (at the very least within the adverts).
I’m in a little bit of restoration mode in the present day, having simply returned from a number of days in northeast Louisiana visiting Meta’s huge AI information heart website, referred to as Hyperion, for a function story I’m reporting.
I’ve been scouring the thesaurus, attempting to land on the precise phrase to explain simply how massive, loud, and chaotic this building website is. Colossal? Mammoth? Sprawling? Let’s put it this fashion: it takes some time simply to drive the size of the location—it stretches roughly 5 miles from prime to backside.
And through that drive, I found that Hyperion is getting even greater. Whereas the growth had lengthy been suspected—and was one thing of an open secret amongst some locals—I confirmed that Meta has quietly bought roughly 1,400 further acres, an space practically twice the dimensions of Manhattan’s Central Park, adjoining to its already-mega 2,250-acre campus. I additionally noticed lively building underway on the newly acquired land—after I wasn’t worrying about getting mowed down by the infinite parade of 18-wheelers hauling supplies in and across the website.
A possible Amazon-OpenAI deal to energy Alexa
However even just lately, beta testers had been voicing loads of frustrations. “When I ask Alexa to turn off the light, it should turn off the light—not everything on the strip,” one software program engineer wrote final October in an inside channel for suggestions on unreleased Alexa+ options. “It turned off the power strip that my aquarium filter is on and killed my fish.”
Different testers complained that the assistant talked nonstop, ignored repeated instructions to be quiet, or blasted music at full quantity when nobody was dwelling.
Alexa’s AI has been a protracted, gradual journey
I’ve been following the journey of Amazon’s Alexa intently because the post-ChatGPT daybreak of generative AI in 2023. That September, Amazon held one other glitzy occasion—this time at its second headquarters in Washington, D.C.—the place David Limp, then the corporate’s head of gadgets and companies, demonstrated a brand new generative-AI-powered Alexa by saying, “Alexa, let’s chat.”
However practically a 12 months later, I reported—primarily based on interviews with greater than a dozen former workers who labored on Alexa’s AI—that the group was beset by structural dysfunction and technological challenges that repeatedly delayed cargo of the brand new generative AI Alexa. These former workers painted an image of an organization lagging behind Massive Tech rivals Google, Microsoft, and Meta within the race to deploy AI chatbots and brokers, and struggling to catch up.
The September 2023 demo, they emphasised, was simply that—a demo. The massive language mannequin on the coronary heart of the brand new Alexa, which Amazon positioned as a rival to OpenAI’s ChatGPT, was, in keeping with former workers, removed from state-of-the-art. Analysis scientists stated Amazon lacked each the information and the specialised computing infrastructure wanted to coach and run modern LLMs at scale.
Since that report, Amazon has shifted to a extra hybrid strategy. Alexa+ has now been powered by a mixture of Amazon’s personal Nova fashions and fashions from Anthropic, the AI startup by which Amazon has invested $8 billion.
And but, issues persist. Can OpenAI actually remedy Alexa’s long-standing issues? Or would an Amazon–OpenAI deal divert OpenAI’s consideration at a second when it’s locked in fierce competitors with Google and Anthropic? And looming over all of it is Apple, whose personal cope with Google to energy Siri complicates an already crowded AI panorama.
Greater than something, the reported talks reveal how determined even the most important gamers have turn out to be to remain forward in a no-brakes AI race.
FORTUNE ON AI
OpenAI declares Frontier, an AI agent platform for enterprises to energy apps like Salesforce and Workday—however might it will definitely change them? – by Sharon Goldman
Unique: Lawhive, a startup utilizing AI to reimagine the final follow regulation agency, raises $60 million in new enterprise capital funding – by Jeremy Kahn
Meta’s Hyperion AI information heart will sprawl to 4 instances the dimensions of Manhattan’s Central Park – by Sharon Goldman
Palmer Luckey says AI will make {hardware} so low cost you’ll have the ability to purchase a ‘Ford F-150 for $1,000’ – by Jake Angelo
Unique: Former Cohere exec Sara Hooker has raised $50 million for her AI startup Adaption Labs—a guess on smaller, smarter fashions – by Jeremy Kahn
AI IN THE NEWS
The AI rout that lastly spooked the market. This Bloomberg piece actually captured the truth that after years of AI-fueled volatility, this week’s market rout stood out for each its velocity and its underlying concern: not that AI is a bubble, however that it’s already starting to interchange whole enterprise fashions. A whole lot of billions of {dollars} had been wiped off international shares and credit score in simply two days, with software program corporations hit hardest, as buyers reacted to indicators that AI instruments are transferring from promise to follow—sparked partially by Anthropic’s new authorized product and bolstered by broader unease throughout the sector. Even AI’s perceived winners confirmed pressure, with Alphabet warning of upper AI spending and Arm Holdings lacking income expectations. Whereas corporations like Salesforce and ServiceNow haven’t but reported dropping prospects to AI-native opponents like Anthropic and OpenAI, disappointing adoption of some in-house instruments—similar to Microsoft’s Copilot—has deepened considerations that AI leaders might overtake SaaS incumbents sooner than anticipated.
Google is on a roll due to AI. A have a look at Google’s last-quarter 2025 outcomes exhibits its AI technique repay. The New York Occasions reported that Gemini fashions have surged forward of rivals, serving to enhance YouTube engagement by way of suggestions and driving extra search visitors by delivering concise, AI-generated solutions—momentum that lifted Alphabet’s quarterly income to $113.8 billion, up 18% 12 months over 12 months, and pushed annual gross sales previous $400 billion for the primary time. Earnings jumped 30% within the December quarter, however the positive aspects include a tradeoff: Google plans to almost double capital spending this 12 months to as a lot as $185 billion because it races to construct AI information facilities to help Gemini at scale, underscoring how even AI’s obvious winners are buying and selling compressed revenue margins for dominance in an more and more capital-intensive race.
Elon Musk is betting his enterprise empire on AI. I loved this text within the Economist, which centered on how Elon Musk is doubling down on AI with a sweeping—and dangerous—restructuring of his enterprise empire. His merger with SpaceX with xAI, valuing the mixed entity at roughly $1.25 trillion, is supposed to offer xAI an edge by pairing it with SpaceX’s launch capabilities and even pursuing the far-fetched thought of orbital information facilities. However the deal additionally ties a extremely worthwhile house enterprise to a cash-burning AI lab that trails rivals like OpenAI, whereas layering on huge debt and regulatory threat inherited from Musk’s social platform X. On the identical time, Musk is remaking Tesla right into a “physical AI” firm centered on robotaxis and humanoid robots—companies which will take years to repay at the same time as Tesla’s core automobile gross sales gradual. The guess is stark: if AI proves as transformative as Musk believes, the gambles might cement his dominance; if not, he could also be stretching even the “Elon backstop” – the concept that he’ll personally step in if one in all his corporations will get into hassle – to its limits.
Hollywood is dropping audiences to AI fatigue. I’m certain this Wired story is touching a nerve, which argues that Hollywood’s obsession with rogue, malevolent AI—from Metropolis to HAL 9000 and Skynet—has warped how the general public understands in the present day’s synthetic intelligence. Reasonably than autonomous superintelligences plotting humanity’s downfall, the actual dangers of AI are way more mundane and rapid: sloppy automation, degraded artistic work, misinformation, and techniques deployed at scale with out accountability. The writer means that these sci-fi tropes distract from how AI is definitely reshaping tradition and labor proper now—not by way of dramatic revolt, however by way of quantity, velocity, and financial strain—fueling viewers backlash towards what many see as low-quality “AI slop” and hole innovation. In different phrases, the hazard isn’t that AI will flip evil; it’s that it’s already reshaping media and creativity in methods folks don’t like and didn’t select.
EYE ON AI NUMBERS
7
That is at the very least what number of Tremendous Bowl adverts this 12 months had been purchased both by AI corporations, corporations hawking AI merchandise, or corporations touting that the adverts themselves had been created with AI.
Along with OpenAI working an advert, its competitor Anthropic is airing its first Tremendous Bowl business about its Claude system and lobbing jabs at OpenAI about its addition of adverts into ChatGPT. Base44, an AI app-building platform, can also be working its first Tremendous Bowl spot.
In the meantime, Meta is pushing its AI-powered glasses in an advert, and Amazon Ring is debuting an advert for an AI-powered function to seek out misplaced canines. Web site builder Wix can also be working an advert for Wix Concord, its AI design platform.
Lastly, vodka model Svedka is touting its Tremendous Bowl advert with largely AI-generated artistic.
AI CALENDAR
Feb. 10-11: AI Motion Summit, New Delhi, India.
Feb. 24-26: Worldwide Affiliation for Protected & Moral AI (IASEAI), UNESCO, Paris, France.
March 2-5: Cell World Congress, Barcelona, Spain.
March 16-19: Nvidia GTC, San Jose, Calif.
April 6-9: HumanX, San Francisco.
