Analyst house IDC this week released its latest quarterly PC-shipments forecast, predicting that demand growth will finally return to the market in 2024, following a couple years of post-pandemic sobriety. Gartner also recently came to a similar conclusion, and it makes sense when you consider that Microsoft will cease to support Windows 10 in 2025, pushing many IT departments towards a refresh.

However, according to IDC, the generative A.I. explosion complicates matters. First, it said in its report, IT decision-makers are now “questioning where to prioritize budgets.” But there is probably also something of a PC revolution just over the horizon, as regular computers take on some of the A.I. tasks that currently require sending data back and forth between the desktop and the cloud—particularly where time-sensitive things like gaming or translation are involved.

Intel CEO Pat Gelsinger is certainly pushing that narrative, saying a month ago in an earnings call that “we see the A.I. PC as a critical inflection point for the PC market over the coming years that will rival the importance of Centrino and Wi-Fi in the early 2000s.” (Intel’s Centrino computing platform was at the time particularly notable for its integrated wireless networking capabilities.) Intel’s Meteor Lake processors, due next month, are its first that are designed to run A.I. applications locally.

But IDC’s analysts are also on board, writing: “While A.I.-capable PCs are not ready today, they are coming and have shifted some of the discussion around device purchasing within businesses.”

Also of note in IDC’s analysis: AMD and Apple both have rising shares of the PC processor market, standing at 11% and 5% respectively last year. The analysts wrote that we could be seeing “some of the biggest shifts in commercial PC history.”

PCs are becoming interesting again—who knew? More news below.

Want to send thoughts or suggestions to Data Sheet? Drop a line here.

David Meyer

NEWSWORTHY

Nvidia soars again. Nvidia’s share price jumped 4.2% yesterday following the announcement of a new Google deal that will give developers access to Nvidia’s A.I.-friendly processors through Google Cloud. According to Reuters, fund managers who thought Nvidia’s stock was too expensive—its price has risen 230% this year—are now being punished.

Apple Watch and iPhone 15. Apple has (as previously rumored) set Sept. 12 as the date for its big annual iPhone reveal. It’s probably going to be called the iPhone 15, and the big new feature will be the introduction of USB-C charging, as mandated by European authorities. New Apple Watches are also expected, and Bloomberg reports there will be new AirPods too, with the case also sporting a USB-C port.

RIP Pixel Pass. Google—which will soon release its iPhone-rivalling Pixel 8—has discontinued the Pixel Pass subscription that it launched a couple years back when the Pixel 6 came out. The U.S.-only subscription bundled together a Pixel phone, device protection, and a bunch of Google services (cloud storage, YouTube Premium, etc.) that otherwise require their own subscriptions. 9to5Google reports that existing subscribers will get to enjoy their services through the end of the two-year contract term, and Google is also giving them $100 in Google Store credit.

SIGNIFICANT FIGURES

$1 billion

—The revenue hurdle that OpenAI will reportedly clear in the next 12 months. This marks a big boost over the firm’s previous revenue projections. Last year’s revenue was just $28 million, but OpenAI wasn’t monetizing ChatGPT yet at that point.

IN CASE YOU MISSED IT

The A.I. industry confronts life after data scraping, by Sage Lazzaro

Why current-generation A.I. is ‘mostly crap,’ but still came along at the ‘perfect time’ to save the flatlining economy, according to a boutique VC firm, by Paolo Confino

Sam Bankman-Fried wants to pay expert witnesses up to $1,200 an hour. Prosecutors are trying to block them, by Leo Schwartz

Meta says it disrupted a massive disinformation campaign from China, by Chris Morris

People are literally dying because it’s so hot outside—turns out machines aren’t doing so great either, by The Conversation

Inside designing a moon-orbiting outpost: Here’s what it takes to make lunar living possible, by Massimo Comparini

BEFORE YOU GO

A.I. insecurity. The U.K.’s National Cyber Security Centre (a division of GCHQ, the country’s equivalent to the U.S. National Security Agency) has published a couple of interesting blog posts about the specific risks faced by organizations that want to incorporate A.I. into their systems.

The first post discusses the cybersecurity implications of adopting large language models (LLMs), specifically prompt injection attacks (using chatbot prompts to trick the system into revealing keys and other sensitive data) and data poisoning attacks (where attackers tamper with training data to “produce undesirable outcomes, both in terms of security and bias.” The second addresses integrations with LLMs, warning of issues such as the models’ inability to “distinguish between an instruction and data provided to help complete the instruction”:

Consider a bank that deploys an ‘LLM assistant' for account holders to ask questions, or give instructions about their finances. An attacker might be able send a user a transaction request, with the transaction reference hiding a prompt injection attack on the LLM. When the user asks the chatbot “am I spending more this month?” the LLM analyses transactions, encounters the malicious transaction and has the attack reprogram it into sending user’s money to the attacker’s account.

Be careful out there!