Dave Lee
My WorkBlogupdated 2 hours ago
ColumnsVideoTVLivestreams

Column: Alphabet Investors Sound a Warning for Profligate AI Spending

Alphabet Inc.’s impressive second-quarter earnings release on Wednesday seemed on course to go down mostly well with investors. Until the earnings call, the Google parent’s share price was up marginally in after-hours trading, buoyed by cloud growth that smashed estimates.

Then executives disclosed their projection that capital expenditures would increase again, sending the stock down by as much as 4.6%. Let it be a warning to the other hyperscalers preparing to report next week: Even with a quarter as good as this one, you’ll still be punished if you announce yet more spending in pursuit of artificial intelligence.

Alphabet Investors Sound a Warning for Profligate AI Spendingbloomberg.comAlphabet Inc.’s impressive second-quarter earnings release on Wednesday seemed on course to go down mostly well with investors. Until the earnings call, the Google parent’s share price was up marginally in after-hours trading, buoyed by cloud growth that smashed estimates.

aialphabetbloombergmarkets

Column: China’s Kimi K3 Strengthens the Case for US AI Investment

Kimi K3 provides a hugely useful foil for those who want to quieten voices calling for a safer, more structured approach to AI development and implementation in the US. The model’s launch bookended a week in which New York state passed a first-of-its-kind moratorium on data center development, a popular move some worry might become contagious in a midterm year.

China’s Kimi K3 Strengthens the Case for US AI Investmentbloomberg.comWall Street has been on alert for any reason to panic over China’s progress on artificial intelligence. But despite some initial ructions, the breakthrough Friday of a Chinese AI startup that the company said rivals models produced by OpenAI and Anthropic wasn't it — with good reason.

aichinakimi

Column: Apple’s OpenAI Suit May Turn Out to Be a Favor

Though we don’t yet know the full merits of Apple’s trade secret lawsuit against OpenAI, we do know one thing: It’s bad news for OpenAI’s efforts to build an iPhone-killer. The legal assault will bog down Sam Altman’s company in injunctions and red tape.

But here’s the reality: OpenAI never really stood a chance of toppling the iPhone anyway. This suit should force the company to give up the folly sooner rather than later, culling yet another dead-end side quest on its way to an initial public offering. In the long run, if OpenAI survives, it might one day be thankful that Apple’s intervention made it change course sooner rather than later, saving billions of its desperately needed dollars.

Apple’s OpenAI Suit May Turn Out to Be a Favorbloomberg.comThough we don’t yet know the full merits of Apple’s trade secret lawsuit against OpenAI, we do know one thing: It’s bad news for OpenAI’s efforts to build an iPhone-killer. The legal assault will bog down Sam Altman’s company in injunctions and red tape.

aiappleopenai

Column: Apple Is Giving Intel’s Turnaround Some Momentum

One of the reasons the [Trump administration's investment in Intel] has so far proved so sound is that the White House has made sure of it. According to The Wall Street Journal, Commerce Secretary Howard Lutnick personally pushed deals on Intel’s behalf with some of the most lucrative clients you could think of. They include Nvidia Corp., the company at the heart of the AI revolution; SpaceX, which would like to be; and now, reportedly, Apple Inc. — a crown jewel client that would be the firmest stamp of approval yet on the Intel turnaround project.

Apple Is Giving Intel’s Turnaround Some Momentumbloomberg.com

Column: Google Is Trying to Integrate Too Much AI Too Quickly

Yet as the I/O keynote flew past the hour-and-a-half mark on Tuesday, it became clear Google faces an issue in rolling out AI to all those services. Hearing the latest announcements was like sitting in a wind tunnel. The danger that consumers will be left overwhelmed, and grow more resistant as a result, is real. In trying to reinvent Google’s services for AI, there’s such a thing as doing way too much way too quickly.

Google Is Trying to Integrate Too Much AI Too Quicklybloomberg.com

Column: Don’t Throw Out the Keyboard in the AI Revolution

Just last week I had a someone reply to a request for an interview using what he referred to as “AI-assisted articulation.” Harshith Vaddiparthy, a tech executive, read my email asking for his view on the reliabilty of Anthropic’s coding tool. Then he spoke into his computer using OpenClaw, a speech-to-text tool, “to create a full doc with my thoughts.” He reviewed it before sending it to me.

“The key nuance I would emphasize is that OpenClaw is not independently inventing answers on my behalf,” Vaddiparthy told me after I quizzed him about it. “The point of view is mine.”

Is it, though? I get what he’s saying: They are his thoughts, structured into text, then approved. But I think he’s kidding himself, as is anyone else delegating to these tools. An unstructured, unarticulated thought spoken aloud is merely a half-thought; a raw egg cracked into a cold pan. “Writing is thinking,” the two-time Pulitzer Prize-winning author David McCullough said in 2002. “To write well is to think clearly. That’s why it’s so hard.”

Don’t Throw Out the Keyboard in the AI Revolutionbloomberg.com

Column: Anthropic’s Capacity Pinch Cracks the Door for OpenAI

That percentage difference might seem negligible, but for demanding workflows, it isn’t. “Anthropic has some fantastic models,” said Wade Foster, CEO of automation company Zapier, who explained that his company switches out models often for a variety or reasons, one of them being reliability. “The whole industry is compute constrained, and Anthropic has procured a lot less compute compared to OpenAI. So yes, I think Anthropic is feeling that.” (Separately, Foster said, Claude Design had failed him just as he was finishing a presentation. “Well, that stinks,” he remembered thinking.)

Anthropic’s Capacity Pinch Cracks the Door for OpenAIbloomberg.com

Column: Apple Is Winning the AI Spending Game by Not Playing It

Separate from the pack is Apple, which has a much different story. Analysts on its Thursday earnings call didn’t bring up capex even once — a contrast to previous quarters when the concern had been that Cook was being too timid in spending compared with Apple’s peers. Instead, focus was on simply whether the company can make its hardware quickly and cost effectively enough to meet what is a surge in demand to use them … for AI.

Apple Is Winning the AI Spending Game by Not Playing Itbloomberg.com

Column: Meta Needs to Stop Spending as If It’s a Cloud Giant

That spending is in the same arena as the other cloud businesses. Except, Meta isn’t a cloud business. So when it breaks the news that it’s going to need to spend even more, it can’t point to healthy growth in cloud sales from AI — because obviously there aren’t any. Investors are right to question what Meta will get in return for all that AI investment when ultimately its only client is itself.

Meta Needs to Stop Spending as If It’s a Cloud Giantbloomberg.com

Column: An OpenAI Bubble Is Not an AI Bubble

I’ve questioned before whether Wall Street has the temperament for the artificial-intelligence era. It’s hard to argue it does when just one report from the Wall Street Journal, which suggested OpenAI had missed some internal growth targets, was enough to wipe billions of dollars of value off related stocks. For a moment on Tuesday, you’d have thought the sky was falling on the AI boom.

An OpenAI Bubble Is Not an AI Bubblebloomberg.com