TechRadar News

• Elon Musk plans AI compute equal to 50 million H100 GPUs within just five years
• xAI’s training target equals 50 ExaFLOPS, but that doesn’t mean 50 million literal GPUs
• Achieving 50 ExaFLOPS with H100s would demand energy equal to 35 nuclear power stations

Elon Musk has shared a bold new milestone for xAI, which is to deploy the equivalent of 50 million H100 class GPUs by 2030.

Framed as a measure of AI training performance, the claim refers to compute capacity, not literal unit count.

Still, even with ongoing advances in AI accelerator hardware, this goal implies extraordinary infrastructure commitments, especially in power and capital.

A massive leap in compute scale, with fewer GPUs than it sounds

In a post on X, Musk stated, "the xAI goal is 50 million in units of H100 equivalent AI compute (but much better power efficiency) online within 5 years."

Each Nvidia H100 AI GPU can deliver around 1,000 TFLOPS in FP16 or BF16, common formats for AI training - and reaching 50 ExaFLOPS using that baseline would theoretically require 50 million H100s.

Although newer architectures such as Blackwell and Rubin dramatically improve performance per chip.

According to performance projections, only about 650,000 GPUs using the future Feynman Ultra architecture may be required to hit the target.

The company has already begun scaling aggressively, and its current Colossus 1 cluster is powered by 200,000 Hopper based H100 and H200 GPUs, plus 30,000 Blackwell based GB200 chips.

A new cluster, Colossus 2, is scheduled to come online soon with over 1 million GPU units, combining 550,000 GB200 and GB300 nodes.

This puts xAI among the most rapid adopters of cutting edge AI writer and model training technologies.

The company probably chose the H100 over the newer H200 because the former remains a well understood reference point in the AI community, widely benchmarked and used in major deployments.

Its consistent FP16 and BF16 throughput makes it a clear unit of measure for longer term planning.

But perhaps the most pressing issue is energy. A 50 ExaFLOPS AI cluster powered by H100 GPUs would require 35GW, enough for 35 nuclear power plants.

Even using the most efficient projected GPUs, such as Feynman Ultra, a 50 ExaFLOPS cluster could require up to 4.685GW of power.

That is more than triple the power usage of xAI’s upcoming Colossus 2. Even with advances in efficiency, scaling energy supply remains a key uncertainty.

In addition, the cost will also be an issue. Based on current pricing, a single Nvidia H100 costs upwards of $25,000.

Using 650,000 next gen GPUs instead could still amount to tens of billions of dollars in hardware alone, not counting interconnect, cooling, facilities, and energy infrastructure.

Ultimately, Musk’s plan for xAI is technically plausible but financially and logistically daunting.

Via TomsHardware

You might also like

• Check out the best 3D modeling software for 3D printing and more
• We've rounded up the best portable monitors available now
• At $250, Dell's cheapest laptop is almost half the price of its most affordable PC

• Eizo FlexScan FLT-S uses just 6 watts of power, making it far more efficient than most office monitors
• Certified as the world’s first Class A monitor under Europe’s updated energy standards
• Connects using just one cable through docking stations, keeping desks clear and clutter-free

Eizo has announced the FlexScan FLT-S, a 23.8 inch full HD monitor that brings energy efficiency, portability, and a clean design into a single display.

Though marketed as a business monitor, its combination of low power draw, simplified setup, and flexible use makes it a compelling alternative to many traditional desktop monitors.

This device is currently available in Japan for 59,950 yen (approx. $406), and it is sold as a made to order product, which suggests that it targets professionals.

Some reasons why this device can replace mainstream systems

At just 6 watts during typical use, the FLT-S offers energy savings far beyond what most standard monitors can match.

Eizo says this makes it the world’s first monitor to earn a Class A energy label under the EU’s 2019/2013 regulation, which reflects its power consumption and efficient internal design.

In office environments with multiple displays running for 8 or more hours a day, this reduction in power use could translate into noticeably lower electricity bills.

It is not just good for sustainability, it is a practical financial benefit, especially at scale.

Compared to larger or older screens that often consume three or four times as much power, the FLT-S shifts the baseline for what users should expect in efficiency.

Another major advantage of the FLT-S is its minimal setup with single cable operation via USB-C or similar docks.

In most workspaces, it can be plugged into a docking station, reducing cable clutter and simplifying desk layouts, which supports a cleaner and more manageable working environment.

Unlike the earlier FlexScan FLT model, which came with a clamping arm, the FLT-S includes a small desk stand that can be placed anywhere without installation, making it suitable for open offices, temporary desks, or situations where mounting is impractical.

It supports tidy, low maintenance setups, which is now expected in modern business environments.

Also, with its lightweight build, simple stand, and plug and play nature, it can be moved around more easily, whether it is a quick desk change or being taken home for hybrid work.

Users do not need to unmount or disassemble anything, just pick it up, take it home, and plug it in.

This mobility places it alongside the best portable monitor options, but with better performance and a full desktop size.

It also means businesses do not have to issue separate screens for home and office. One unit does both jobs well.

Via mynavi

You may also like

• We've rounded up the best monitors for photo editing on the market today
• And here are the best monitors for programming we've found
• Samsung's largest SSD to date goes on sale for $5,593 — 61.44TB PCIe Gen5 SSD costs only $0.09/GB

Apple quarterly earnings reports are not usually fodder for consumer interest. It's a lot of sales numbers and explanations about why the falling dollar, rising yen, or China headwinds are impacting sales and earnings results.

Without the benefit of product introductions, it can be a snoozefest, but this time should be different, and mostly because of Apple's not-so-great year.

• To recap, Apple had to delay the fully Apple Intelligence-upgraded Siri and admit that what they had still wasn't up to Apple's standards.
• The company appears to be falling behind in the critical AI race and has yet to offer a solid, competitive answer.
• It had to shuffle executives to try to get Siri and Apple Intelligence back on track.
• While we wait for a folding iPhone, Samsung just released what may be the industry's best flexible phone.
• The Vision Pro remains an expensive niche product that has not inspired vast numbers of consumer customers.
• The spectre of tariffs is putting pressure on Apple and its partners and may soon force it to start charging more for tech gear.
• The company has been pressured by the Trump administration to start building iPhones in the US.

Put simply, Apple has a lot of explaining to do. So while Apple CFO Kevan Parekh and CEO Tim Cook will spend the first half of the earnings call that is set for Thursday, July 31 at 5PM ET (2PM PT / 10PM BST), droning through profits, operating expenses, capital, and even stock splits, it's the open call with analysts that should be most illuminating, important, and, possibly, even entertaining part of the event.

Analysts should be quizzing Cook and company on these juicy topics:

AI and Apple Intelligence

As I noted above, Apple has fallen far behind in the all-important AI race, and promises that the updated Siri will show up "in the coming year" are less than comforting, especially since that might be next year.

Analysts will rightly demand specifics. And while I do not expect Cook to get pinned down, he might offer some assurance that a Siri that can compete with, say, Gemini or ChatGPT, will arrive by the end of this year, as opposed to slipping into 2026.

I would love to see analysts press Cook on Apple's overall AI strategy, one that I believe is flawed for being too cautious. The competition is flying down AI Highway with abandon, and with Superintelligence or General Artificial Intelligence on the horizon, Apple cannot afford to take the slow approach. Any more delays and Apple will lose more than just this AI race.

New Products

During every earnings call, analysts make desperate attempts to get Cook to mention upcoming products. He never does, but Cook will talk vaguely about "the best lineup ever" of upcoming technologies. His enthusiasm can often speak volumes about what to expect and if any of it will move the needle.

(Image credit: Shutterstock ID 1870904317)Vision Pro pump up

Even though Vision Pro fails to dazzle on the sales side, it remains the most powerful and perhaps the best consumer technology Apple has ever produced. I expect Cook to highlight consumer and enterprise interest, as well as recent content successes, such as the Bono Documentary.

It's the analysts' job, though, to press Cook here and see if they can get him to admit that Vision Pro will never be a consumer product, at least not at its current price.

AR embrace (iGlasses, anyone?)

A good segue here would be a return for Cook to mentions of an AR future. Apple's wearable game cannot remain confined to watches and earbuds, not when Meta is making hay with all those Meta Smart Glasses from Ray-Ban and Oakley.

If we only consider AR glasses, Apple still has some time since Google, Samsung, and Meta are all still trying to figure out how to make high-quality lenses that do not need the bulk of larger frames to support them.

Could analysts goad Cook into mentioning future "iGlasses"?

Airy or bendy phone possibilities

Most people expect Apple to deliver its thinnest iPhone ever this September in the form of an iPhone 17 Air. Cook will not name this product, but he could mention "new form factors," which could be referring to the thinner Air and, maybe, a folding iPhone.

In both areas, though, Apple is behind Samsung, which now has the best and thinnest foldable design in the Samsung Galaxy Z Fold 7 and an admirably thin, if uninspiring, Samsung Galaxy S25 Edge.

Analysts will want at least a hint that Apple has an answer for all this.

Tariffs and US-based manufacturing

Cook will not want to talk about US-based manufacturing, tariffs, or the guy who calls him a friend, President Trump. But analysts will ask and press for projections on how tariffs might impact iPhone and other Apple gadget pricing.

Here, I expect Cook to offer at least some color, if not concrete projections. He'll talk again about how Apple is prepared for supply chain fluctuations, which include component pricing pressure. He will assure everyone that Apple has a plan for this uncertain future.

Cook might remind people about how Apple has already diversified manufacturing so that it's not all in China and point to the $500 billion investment in the US, which most recently has included manufacturing training programs for upcoming businesspeople and their businesses.

I don't know if that will satisfy everyone, especially not Trump, who has consistently demanded that Apple build the iPhone in the US.

At least Cook will get to tout the sea changes coming to all of Apple's major platforms and the impact of Liquid Glass on, for instance, the iPhone. iOS 26, iPadOS 26, and macOS 26 are big updates and ones that, whether or not people love them, do promise to change how they use these platforms.

Overall, this could be one very exciting and even contentious earnings report, and I'll have my popcorn ready.

You might also like

• I've been using iOS 26 for a month – here are 3 things I ...
• How to download the iOS 26 public beta
• iOS 26 public beta is now live – here are the 5 biggest new ...