Nearly a year ago, OpenAI released ChatGPT 3 into the world, and investors got visions of dollar signs in their heads as they imagined the ways that artificial intelligence could make big money for businesses.
Wall Street’s now coming to terms with the fact that those sorts of paydays are going to take time. As investors have already seen from the past two quarters of earnings, AI has only really delivered financial benefits for a select few hardware companies so far — while spurring new costs for many others.
“The AI boom has already bifurcated into the contenders and pretenders,” said Daniel Newman, chief executive and principal analyst of Futurum Research. And while Advanced Micro Devices Inc., Intel Corp. and Arm Holdings PLC
ARM,
have stirred up interest, Nvidia Corp.
NVDA,
has established itself as far and away the greatest “contender,” with AI driving strong demand for its chips tuned for AI training.
Nvidia last quarter reported record earnings, including a 141% jump in revenue for its graphics chips used in AI infrastructure building up data centers. Nvidia, which reports near the end of earnings season on Nov. 21, posted record revenue of $13.5 billion last quarter and is expected to easily top that with $16 billion in the most recent quarter, a surge of 170% versus a year ago. Those estimates include $12.3 billion of revenue coming from data-center sales.
Other chip companies could post gains from AI as well, but to far lesser extents. Candidates include Broadcom Corp.
AVGO,
and system maker Super Micro Computer Inc.
SMCI,
as well as Marvell Technology Inc.
MRVL,
which last quarter told analysts that it expects to end the year at a revenue run rate of about $800 million this year from cloud/data-center chips related to AI.
“This is well above what we had outlined last quarter. Put this in perspective: This would put us at the run rate we had previously communicated for all of next year,” Marvel Chief Executive Matthew Murphy told analysts.
Super Micro is also riding the AI wave with its customized data-center servers that are designed to consume less power. But revenue in the September quarter is forecast to rise just 15% from a year ago and drop on a sequential basis, as supply constraints from Nvidia likely hampered Super Micro’s ability to meet all its demand.
Much as Advanced Micro Devices Inc.
AMD,
and Intel Corp.
INTC,
want to be in the AI conversations with the graphics chips they hope will be used for AI data-center applications, they won’t see much of an impact yet from AI revenue. Plus, those companies are experiencing a slowdown in PC sales that may overshadow any small benefit from AI chips.
The AI boom in chips is clearly not providing enough of a boost to lift finances for the overall semiconductor sector, which is forecast to see earnings fall 3.3% in the third quarter and post a revenue decline of 0.6%, according to FactSet. The industry is being dragged down in part by Micron Technology Inc.
MU,
which reported a 40% drop in revenue and a whopping fiscal fourth-quarter loss in late September for the quarter ended Aug. 31, which is included in FactSet’s third-quarter data. Even so, the company called a bottom to the memory-chip downturn.
Read also: Micron’s AI focused chip won’t help financial results anytime soon.
“Most of the consumer-based tech is still struggling, [including] PCs, laptops and to a certain extent smartphones,” said Daniel Morgan, senior portfolio manager at Synovus Trust Co. Wall Street has tempered expectations related to the impact of Apple Inc.’s
AAPL,
iPhone 15 launch on the quarter, as estimates call for an overall 1% drop in September-quarter revenue. Last quarter, Apple executives forecast that both Mac and iPad sales would be down by double-digits and that revenue performance would be similar to its June quarter, when revenue fell 1.3%
In addition, when asked about AI, Apple CEO Tim Cook said the company views AI and machine learning “as core fundamental technologies that are integral to virtually every product that we build.” Those comments, though, can also apply to the bulk of tech companies, where AI is built into software as another layer to improve a product. Internet companies such as Meta Platforms Inc.
META,
and Alphabet Inc.
GOOG,
GOOGL,
incorporate AI into their software and algorithms but don’t treat it as a specific, revenue-generating product.
Other software companies are building AI into their products as separate features or add-ons, but they are still in the early stages of seeing whether or not customers will pay more for them. Take Microsoft Corp.,
MSFT,
which has showed off Copilot, an extra AI feature for customers of Microsoft 365.
“[Microsoft] can distinguish itself by providing more details around its AI revenue
ramp since we don’t expect much information from Google, who really doesn’t seem
to have the monetization plan for Bard and AI-assisted search (SGE) ready to
articulate yet,” Melius Research analyst Ben Reitzes said in a note to clients this week. He also noted that the cost of offering AI products to consumers is steep, and requires lots of investment.
“There are sophisticated issues to contend with for Microsoft, including balancing the potential for higher revenue from Copilots with the high costs per query and much-needed investment,” Reitzes said. “The balance of AI adoption vs. cost was implied when Microsoft guided to flat operating margins year over year for fiscal 2024.”
Earlier this year, the Information reported that OpenAI, the creator of ChatGPT and recipient of a hefty investment from Microsoft, has costs of up to $700,000 a day, because the massive amounts of computing power needed to run queries. In February, OpenAI launched ChatGPT Plus, for $20 a month, a service that will give subscribers access to its AI during peak times and faster response times.
Another example is Adobe Inc.
ADBE,
which has a few AI offerings, including a subscription service called Generative Credits, tokens that let customers turn text-based prompts into images. Another is Firefly, a generative AI service for images, and an AI option in Photoshop, currently called Photoshop Beta AI, to help users fill in images and other collaborative tools. Adobe did not provide any forecasts on potential revenue generation during its analyst day earlier this month.
Toni Sacconaghi, a Bernstein Research analyst, said AI could drive a massive increase in enterprise productivity, and companies could dramatically increase IT spending on servers in order to invest in productivity-enhancing AI. “However, we note that enterprise adoption appears to be in early stages,” he said in a recent note to clients, adding that it was feasible that spending on AI infrastructure could take money away from other IT projects in process. “We do worry that projected AI infrastructure build out may be occurring too quickly, necessitating a digestion period, which could result in a commensurate stock pullback in AI-related names.”
Overall, the information-technology sector itself is expected to see anemic revenue growth this quarter. The consensus on FactSet forecasts a meager 1.35% revenue uptick in the third quarter, with earnings growth of 4.65%. FactSet’s estimates for IT companies exclude internet companies like Meta and Alphabet, which are under the category of communications/interactive media services. That sector is expected to see sales growth of 12%, and earnings growth of 51%, thanks to a 116% boost in Meta’s net income, after it hit a low point in the year-ago quarter.
Amazon.com Inc.
AMZN,
in the category of consumer discretionary/broadline retail, is forecast to see earnings growth of 109%, and revenue growth of 11%. Amazon’s cloud services business, AWS, is expected to also see a potential uplift from customers spending money on AI projects, according to a TD Cowen & Co. survey, in which 41% of respondents said they were “highly considering” allocating a budget for generative AI.
“This trend could bode well for Amazon’s AWS,” TD Cowen analyst John Blackledge said in a recent report, adding that he expects AWS revenue growth to reaccelerate in the second half of this year and in 2024, boosted by the move of additional workloads to the cloud, possibly including generative AI.
As companies build up their infrastructure, or their spending on cloud computing to add or improve AI capabilities, they are seeing higher costs, which is affecting margins — especially if revenue has slowed down, as it has in some sectors. Across both the broader S&P 500
SPX,
and the IT sector, earnings are lower than a year ago.
As Newman of Futurum pointed out, “AI stole the budget this year.” And that is a mixed bag for tech.
