🤦 RTO? RT-No!

With media giants Paramount, AT&T, and The New York Times joining Microsoft and Amazon in stepping up their office attendance requirements, Corporate America seems keen to return back to the old normal... if only their employees would heed the call.

• A growing number of return-or-exit ultimatums and crackdowns from companies don’t seem to be moving the needle, as the share of time that Americans spend working from home has plateaued for much of the last year. 

• Data first reported by The Wall Street Journal from the US Survey of Working Arrangements and Attitudes reveals that an average staffer has been spending about a quarter of their working time from home since 2023, when the share gradually dropped from a pandemic peak of 62%.

• The percentage of full paid days worked from home has bounced between 25% and 30% for essentially four years at this point. 

Per the WSJ, people also chose to spend merely 1% more days in the office every month since the start of 2024 — about 1.4 hours more on average — even as 12% more companies in the US began requiring employees to work in the office over the same period.

That said, things might be changing. The Federal Reserve’s August labor market update observed that some employers were “reducing head counts through attrition — encouraged, at times, by return-to-office policies,” as working from home increasingly becomes a nonnegotiable for many, even if that means voluntarily quitting or cutting pay at times.

The Takeaway

Your typical worker is putting in between one and two days worked from home per week. Companies have tried to squeeze workers a little bit to get that down to, say, zero days per week, but we appear to have hit something of a national impasse at this point. Workers really seem reluctant to give up that one day from home, and companies haven’t really been able to coax that last day out of them by any means outside of outright termination. Where do things go from here? It depends on that labor market. 

Read more.

Hyperion, Colossus, Prometheus, Stargate — these are the names of movies, mythical heroes, books… and massive AI data centers. And there’s a new yet unnamed player entering the arena as Nvidia and OpenAI’s $100 billion megadeal sets a new threshold for unprecedented computing power.

It’s hard to keep up with the AI data center news, so Sherwood News’ tech reporter Jon Keegan broke down the race for gigawatts: 

• Nvidia and OpenAI’s $100 billion project: The plan is to build a staggering 10 gigawatts of computing power supported by Nvidia GPUs — 4 million to 5 million of them.

• Project Stargate: The Texas-based project will produce a 4.5-gigawatt AI data center and was initially slated to cost $500 billion, though plans may be scaled back.

• Hyperion: Meta’s city-sized data center in Louisiana will scale up to produce 5 gigawatts and cost $10 billion... or $50 billion, depending on who you ask.

• Prometheus: Another data center from Meta, but in Ohio, this is the first of the “titan clusters” that hope to achieve “superintelligence.”

• Colossus and Colossus II: Elon Musk’s first xAI data center, Colossus, was built in a record 122 days in Tennessee and has 230,000 GPUs. A second Colossus in the same location is under construction and will have almost double the number of GPUs.

• Project Rainier: Amazon and Anthropic’s site is an “ultracluster” of 30 data centers in Indiana powered by Amazon’s custom Trainium 2 chips.

• Fairwater: The most mellowly named of the bunch, Microsoft’s Wisconsin data center still packs a huge punch with “hundreds of thousands” of Nvidia GPUs.

Keegan has plenty more detail if all that data hasn’t already broken your circuits.

The Takeaway

While only Meta CEO Mark Zuckerberg has said the quiet part out loud, conceding that AI might be a bubble but that “misspending a couple of hundred billion dollars” is worth it to achieve superintelligence and avoid being left behind in AI (*cough* Apple), it’s not hard to see these titanic data centers as a modern-day arms race. Added up, it could be close to a trillion dollars being poured into the battle and hoping the returns are worth it. 

Read more.

Why Nvidia almost has to invest in OpenAI and Intel

Bank of America analyst Vivek Arya breaks it down: “Investing in other public assets has become difficult,” he wrote, adding that the only alternative “is to invest in the ecosystem to expand the size of the addressable opportunity that could multiply future benefits, or accelerate time to market for new products.” But there’s another reason that is particularly germane to Intel. 

Go deeper.

• IonQ jumped after announcing a tech breakthrough on interconnecting quantum computers

• Firefly Aerospace, a Texas-based midcap builder of space launch vehicles, dove after posting its first quarterly report since its August IPO

• Boeing rose on an $8.5 billion deal with Uzbekistan Airways that could be followed by more large orders

• TV station owner Sinclair climbed following news that it won’t air Tuesday’s return of “Jimmy Kimmel Live!”

• Micron reported strong Q4 results and offered earnings guidance above every Wall Street analyst’s estimates

• Meta has launched a federal super PAC to fight state AI policy proposals

• Dogecoin struggled despite the recent doge ETF launch and more upcoming funds

• Disney+ subscribers are getting (another) price hike 

• Why Nvidia’s investment OpenAI is a “watershed moment,” according to analyst Dan Ives