The Bastard Art of Thinking like a Machine
When confused by technology, it might help — and cost less — to switch back to human-mode
Opinion
Microsoft, Data Centers and the AI Paradox
Taking a page from Bitcoin miners constantly seeking cheaper power sources, Microsoft Corp.’s planned $3.3 billion investment in Wisconsin represents the next wave of data center expansion driven by generative AI.
The massive project, which is also being promoted as a jobs creator, is nevertheless bound to strain power grids already groaning under the weight of surging electricity demand.
With data center hubs like Northern Virginia approaching full capacity, and quickly running short of power, Microsoft and other hyperscale cloud providers are eyeing new locations away from the coasts as electricity demand soars. The rise of generative AI and the resulting explosion of new server farms required to handle large-language models are expected to drive demand for electricity through the roof.
Read More »Microsoft, Data Centers and the AI Paradox
Can TSMC Turn Arizona into Taiwan?
What’s at stake:
TSMC is in a box in Arizona. It needs to explain its situation more frankly and transparently. If it doesn’t, TSMC looks more and more as though it plans to “make Arizona Taiwanese.”
A statement by TSMC CEO C.C. Wei at its latest earnings call raised eyebrows among semiconductor industry veterans. Wei said the cost to manufacture chips in Arizona will exceed chips made in Taiwan. TSMC expects customers to share the burden of the price hike.
As my colleague Bolaji Ojo wrote last week, many industry observers wonder how this prospect might fly with TSMC clients like Apple and Nvidia.
Before these big customers bite the TSMC bullet, they want to know more about long-term benefits it is receiving from federal, state and local governments.
More important, everyone deserves more specifics from TSMC as to how it plans to succeed in Arizona. Asking customers to share the higher cost of doing business in the US is one thing. But by blaming its thus far clunky operations in Arizona on American “work habits,” TSMC risks losing the trust of its generous host nation.
Read More »Can TSMC Turn Arizona into Taiwan?
Generative AI: The Bloom Is Off the Rose
Artificial Intelligence is not living up to the hype. Hallucinations are an intractable part of the problem.
AI: Should I Be Faithful or Agnostic?
Artificial intelligence is progressing at a pace that would make Gordon Moore dizzy.
Pick any headline. From the unveiling of Nvidia’s Blackwell platform, supposedly heralding the arrival of trillion-parameter-scale AI models, to Open AI’s highly anticipated release of Chat GPT 5 (mid-2024) and Elon Musk’s latest Tesla robotaxi iteration (to be unveiled on August 8), we are swept into a barrage of promises and proclamations of the progress of artificial intelligence technologies.
Read More »AI: Should I Be Faithful or Agnostic?
Smart vs. Useful
What’s at stake:
Companies need to stop applying the shopworn adjective “smart” to every new chip, system, apps or high-tech gadget. Instead of smart for smart’s sake, how about we focus on “useful” for the user’s sake?
New devices are almost inevitably “smart” in the eyes of their inventors, but are the rest of us too dumb to appreciate them? Or maybe the problem is not “us” at all? These are questions the market – investors, partners, developers, system designers, users and consumers – should be asking.
For too long, the tech industry has been heralding anything new they pump into the market “smart.” What I see too often, however, is “smartwashing,” using claims of smartness to mask the complexities of implementing the technology and to dismiss challenges that the technology imposes on developers and users.
If smart stuff don’t perform like clockwork all the time, the alibi goes, well, the system designers messed up. Or it was consumers who didn’t read the fine print in the users manual.
Read More »Smart vs. Useful
Where Nvidia’s Huang Went Wrong on Thomas Edison vs. AI
What’s at stake:
In Silicon Valley, power prevails. In the short run, if your business and technology have the power to make money, move the market, and change the rules, you won’t even be criticized when you’re wrong. But no business is future-proof, especially if its power depends on a resource — namely electricity — facing increased scarcity.
The Bay Area just finished a week-long AI party, talking up advancements like Nvidia’s new supersized AI chips and the simulation and software tools, developed by companies such as Ansys and Synopsys, that enabled them.
The tech community is blithely comparing the anticipated spread of AI to electricity that everyone nowadays takes for granted.
Executives like to style themselves as creators of a brand-new industry, similar to the electrification wave that accelerated the Industrial Revolution.
Read More »Where Nvidia’s Huang Went Wrong on Thomas Edison vs. AI
What Will Nvidia Do for an Encore?
By Bolaji Ojo
What’s at stake?
Nvidia Corp.’s stratospheric sales growth and lofty stock valuation cannot mask its vulnerability to the semiconductor industry’s roiling and disruptive dynamics. The company now has targets on its back from rivals and even some customers smarting from the elevated pricing of its products. Can Nvidia string together a multi-year, unbroken grip on the AI market and break new ground to maintain its enviable performance?
Jensen Huang should be getting tired of posing for pictures.
At GTC2024, Nvidia’s GPU technology conference in San Jose, Calif., Nvidia’s founder and CEO, Monday, delivered his keynote speech in a rock-concert setting to an audience of more than 10,000 people — a record turnout for the annual event. Enthralled by Huang, many of them, including trade and mainstream journalists, made a beeline for the keynote and giddily took selfies with him.
There is no bigger celebrity in the semiconductor world today than Huang. Even other CEOs in the electronics industry are enamored with him.
“He is a giant-rock star,” said Sassine Ghazi, president and CEO of electronic design automation (EDA) and semiconductor IP supplier Synopsys Inc., speaking Wednesday at SNUG (Synopsys User Group) in Santa Clara, his company’s annual event for chip designers. Huang, an invited speaker at SNUG, swiftly returned the compliment when he joined Ghazi on the stage. “Synopsys is the most consequential company in our industry today,” Huang said. “Without Synopsys tools we will not be able to do what we do.”
Read More »What Will Nvidia Do for an Encore?
The Unfulfilled Promise of Silicon Carbide
The power semiconductor material silicon carbide seemed poised for a different sort of band-gap leap in recent weeks as developers again touted SiC’s potential role in addressing skyrocketing electricity demand. That demand will only grow as sprawling data centers take on more energy-intensive AI workloads.
We were expecting to hear more about the promise of SiC technology and new applications this week from a key developer, Onsemi. Hours before we were to be briefed on its strategy, the chip maker based in Scottsdale, Ariz., abruptly postponed its announcement that appeared to be tied to a co-located event with AI chip giant Nvidia. No word on when Onsemi will reschedule its SiC platform announcement.
Read More »The Unfulfilled Promise of Silicon Carbide
MicroLED Displays Can Survive the Apple Blow
By Peter Clarke
Late last month, Ams Osram AG (Unterpremstatten, Austria) sent out a press release saying that “a cornerstone project underpinning its microLED strategy got unexpectedly cancelled today,” thus prompting the company “to reassess its microLED strategy.” That triggered an avalanche of speculations that Apple may have just killed off the prospects for microLED technology.
I argue that Apple’s move has not killed those prospects off entirely. Ams Osram and a number of microLED startups have been undermined by Apple’s change of mind and must now adjust to a slower-moving environment.
Read More »MicroLED Displays Can Survive the Apple Blow