Artificial intelligence's rise is already beginning to reshape tech's business landscape, with the emergence over the past week of the first big winner: chipmaker Nvidia.

Driving the news: Nvidia opened trading Tuesday with a market capitalization that topped $1 trillion before closing just shy of that level. That placed it in an elite club that also includes Apple, Google/Alphabet, Amazon and Microsoft.

Of note: The jump followed Nvidia's announcement this weekend of several new processors and systems optimized for AI-related tasks.

• CEO Jensen Huang said at an event in Taiwan that the firm's GH-200 Grace Hopper "Superchip" — a module that's tuned for generative AI uses — is now in full production.
• 256 of those "superchips," in turn, power a new AI-focused supercomputer Nvidia rolled out at the same time.
• Such products can help speed up the lengthy, computationally demanding work of training and running large language models, like OpenAI's GPT-4.

How it works: Large language models really are huge — and getting huger.

• They digest trillions of words and boil them down to billions of "parameters" — numbers that represent vast expanses of interrelationships among those bits of language.
• That's why the current wave of generative AI pushes the boundaries of today's hardware capacity in both processing and memory.

Chip companies with products like Nvidia's, which deliver the best performance on the kinds of tasks AI demands, are among the clearest initial beneficiaries of the AI boom.

• Nvidia has also pioneered software tools that combine with its chips to provide a fast foundation for training and deploying the large language models that run the latest AI services, like ChatGPT.

Between the lines: At the onset of a new era, it's impossible to know who will strike long-term gold. But, as was the case in 1849, it's hard to go wrong in the short term by focusing on those with the best pick-axes.

• Microsoft is shaping up as an early winner thanks to its partnership with OpenAI. Microsoft has added AI-powered copilots to all its major software franchises and is also hoping the technology will revitalize its effort to take on Google in search.
• Google is also a leader in AI and could well emerge as a beneficiary of the trend. At the same time, the use of generative AI to deliver conversational answers could upend the search-ad business that drives the bulk of its revenue and profits, meaning the company may have as much to lose as gain.

The big picture: The arrival of a major new platform always reshuffles tech's deck of winners and losers.

• Qualcomm, for example, was a significant player in cellular technology, but became a major global player thanks to the smartphone revolution.
• Likewise, Intel dates back to the late 1960s, but saw its star really rise with the arrival of the personal computer in the 1980s — and then it truly hit the stratosphere when the internet took off in the '90s.

Be smart: It's not always that companies see the future coming.

• Nvidia, for example, developed its core technology to power PC graphics cards.
• The approach, however, turned out to be good for a whole bunch of other purposes, including cryptocurrency mining — and now AI.

Yes, but: Such shifts also lead some players to decline in value and influence.

• IBM remains a large technology company, but doesn't lead the industry the way it did in the mainframe era.
• Intel has also suffered, first by missing out on mobile and, more recently, from manufacturing missteps — along with the emergence of new kinds of computing tasks that its products don't always perform best at.
• The combination has created quite the "mud hole" to dig out of, as Intel CEO Pat Gelsinger acknowledged in an interview in Tuesday's Wall Street Journal.

The FCC, wrangling a first-of-its-kind challenge process, on Tuesday released the most accurate map yet of who in the U.S. has internet access, Axios Pro's Maria Curi reports.

Why it matters: The accuracy of the FCC's broadband map will determine how successful the Biden administration and Congress are at achieving their $42 billion goal of ensuring every American has access to affordable, high-speed internet.

• Consumers, states, localities, tribes and others are able to dispute whether the map is accurately depicting the internet service they get.
• This is the first version of the map that reflects those challenges.

Flashback: The FCC's previous practice of creating broadband availability maps based on census blocks led to an overstatement of who had access.

• If there was service in a single location, the agency assumed there was service in the entire block.
• Census block-based maps identified just 8.1 million locations where internet could be installed.

By the numbers: Under today's more granular mapping effort, the FCC has so far identified more than 114 million locations where internet could be installed.

• 330,000 more unserved locations have been identified since the FCC came out with the first iteration of the broadband map in November that had not yet been subject to any challenges.
• More than 75% of the 4 million+ challenges to the map have been resolved, according to the FCC.

What we're watching: The FCC is investigating whether internet providers are exaggerating their coverage after local authorities accused companies of overstating their service and lawmakers applied pressure.

• FCC chair Jessica Rosenworcel in a Feb. 3 letter to lawmakers said the agency expects the challenge process will serve as a deterrent to over-reporting.
• "We recognize also that as providers gain familiarity with this system, efforts to intentionally misstate service may be subject to enforcement action. In fact, we already have an investigation underway," Rosenworcel said.

What they're saying: NTIA said in a blog post that the agency, which is in charge of working with states to implement broadband grants, "is confident that with this data as a baseline, we will be able to effectively allocate funds by the end of June."

What's next: Major updates to the maps will be released twice a year.

A year ago, nobody was talking about generative AI. Now, it's all anyone wants to talk about — including on company earnings calls.

By the numbers: Mentions of "generative AI" in U.S. corporate earnings calls were up 308% from the previous quarter, per market intelligence firm AlphaSense.

• Generative AI was mentioned in more than 232 earnings calls in May alone — compared to just five mentions of the term during all of 2022.

On Tap

• Salesforce and Crowdstrike report earnings.
• Computex continues in Taiwan.

Trading Places

• Chipmaker Semtech named industry veteran Paul Pickle as its next CEO.

ICYMI

• Block Party says it is putting its anti-harassment tool for Twitter on "indefinite hiatus," citing changes to that company's API policy.
• Google is said to be planning a second Pixel Watch for later this year featuring a Qualcomm chip — rather than the Samsung processor used in last year's model — and more sensors. (9to5Google)

A couple important Lego updates for you. First off, I built a Lego version of the 7-Eleven near my house where I get a Super Big Gulp each day.

• And, perhaps of broader appeal, Lego has announced a new Icons of Play soccer set featuring several female global soccer stars, including Megan Rapinoe and Sam Kerr.
• The 899-piece set is part of a broader #PlayUnstoppable initiative aimed at showing girls that they can play, and do, whatever they want.