Artificial intelligence and generative AI remain the proverbial hype trains of thematic investing this year. With AI-centric companies like NVIDIA carrying major indexes ever higher on outsized earnings beats quarter after quarter, it’s no wonder investors flock to all things AI. But is the hype warranted? Goldman Sachs alongside a number of experts recently weighed in, saying yes… and no.
A Generative AI Overview
Let’s start with a look at what generative AI actually is. While artificial intelligence spans a range of capabilities and types, generative AI garners the most attention. It’s the AI programs that produce images, text, and other content in response to prompts. It uses deep learning models trained on a bank of data to offer predictive answers. Where that data is sourced from remains a highly contentious issue, but many generative AI programs scour the internet or collated pools of training data as sources.
Generative AI in its current iteration faces a number of notable limitations. From concerns over bias and overtly incorrect responses to outright plagiarism, the use of generative AI remains an ethical and legal quandary for many businesses. While it utilizes predictive outputs to formulate responses, it remains incapable of original ideation in its current form.
For investors though, the development and proliferation of generative AI heralds a new era of technology and investing potential. NVIDIA, the enormously popular chip maker and platform developer, remains the golden child of the AI investing story. With the company’s stock up nearly 159% year-to-date on a price basis as of 07/15/24 according to Y-Charts, the company continues to roll out aggressive earnings forecasts each quarter on AI demand.
Into the mix stepped Goldman Sachs last week with a comprehensive gen AI outlook. The research piece included both doubters and heralds of generative AI. With over $1 trillion in capex forecast to be spent on generative AI in the upcoming years, it’s worth attempting to look holistically at the full potential and implications of the technology.
Optimism Abounds for Gen AI’s Potential
Generative AI optimism is well represented amongst those at Goldman Sachs. From the economic upside potential to the differences in the current capex cycle to historical ones, positive perspectives abound. Here’s snippets of what the proponents had to say.
“We have long argued that generative AI could lead to significant economic upside, primarily owing to its ability to automate a larger share of work tasks,” explained Joseph Briggs, Senior Global Economist at Goldman Sachs. This leads to “our baseline estimate implying as much as 15% cumulative gross upside to US labor productivity and GDP growth following widespread adoption of the technology.”
Image source: Goldman Sachs
Those benefits extend beyond borders too. “For European utilities, an industry with elevated operational and financial gearing, the coming inflection in power demand should have significant positive implications for revenues, and, in turn, profits.” That’s according to Alberto Gandolfi, Head of European Utilities Equity Research at Goldman Sachs.
AI Still Has “Room to Run”
Ryan Hammond, senior U.S. Equity Strategist at Goldman Sachs, sees room for growth despite recent high valuations. “More broadly, we believe the AI theme has room to run, with scope for its beneficiaries to broaden as investors look to the next phase of the AI trade, and think this will benefit Utilities in particular.”
“Those who argue that this is a phase of irrational exuberance focus on the large amounts of dollars being spent today relative to two previous large capex cycles,” explained Eric Sheridan, Senior Equity Research Analyst at Goldman Sachs. These include the period of the rise of the desktop computer as well as the cycle of 5g expansion, smartphone uptake, and more.
Instead, investors should look to dollars spent compared to company revenues according to Sheridan. Current “levels are not materially different from those of prior investment cycles that spurred shifts in enterprise and consumer computing habits.”
Kash Rangan, Senior Equity Research Analyst at Goldman Sachs, also weighed in on the AI capex cycle. “Spending is certainly high today in absolute dollar terms,” said Rangan. “But this capex cycle seems more promising than even previous capex cycles because incumbents — rather than upstarts — are leading it, which lowers the risk that technology doesn’t become mainstream.”
Elevated Levels of Generative AI Skepticism
Opposite the supporters are those more cautiously positioned regarding generative AI’s actual potential. A reoccurring theme within the report are the significant hurdles to the proliferation of gen AI. This comes in the form of energy infrastructure constraints, AI infrastructure limitations, and more.
Allison Nathan, Editor at Goldman Sachs, interviewed Daron Acemoglu, Institute Professor at MIT. Acemoglu currently forecasts for muted impacts to labor productivity and GDP estimates in the next 10 years. This expectation is based on the low likelihood of meaningful “truly transformative changes” in the coming decade. It also relies on reduced expectations regarding the rate of AI evolution.
“But too much optimism and hype may lead to the premature use of technologies that are not yet ready for prime time,” Acemoglu cautioned. “Too much automation too soon could create bottlenecks and other problems for firms that no longer have the flexibility and trouble-shooting capabilities that human capital provides.”
Acemoglu isn’t alone in this sort of cautious perspective. Jim Covello, Head of Global Equity Research at Goldman Sachs, casts doubt on the ability of AI to ever solve complex problems. It’s a problem with a $1 trillion price tag.
“So, the crucial question is: What $1tn problem will AI solve? Replacing low-wage jobs with tremendously costly technology is basically the polar opposite of the prior technology transitions I’ve witnessed in my thirty years of closely following the tech industry.” Covello also casts doubt on AI’s future adoption curve, citing limitations to AI cognitive reasoning.
Energy Infrastructure Constraint Realities
A very real challenge that broad AI adoption faces is that of energy and grid capabilities. Brian Janous, Co-founder of Cloverleaf Infrastructure, a developer that helps utility companies generate new grid capacity, discussed the growing divergence of supply and demand.
“Utilities have not experienced a period of load growth in almost two decades and are not prepared for — or even capable of matching — the speed at which AI technology is developing,” explained Janous. “The US has unfortunately lost the ability to build large infrastructure projects — this is a task better suited for 1930s America, not 2030s America.”
Image source: Goldman Sachs
Carly Davenport, Senior U.S. Utilities Equity Research Analyst at Goldman Sachs, translated increased energy demand into hard data estimates. Davenport forecasts data centers to double their electricity usage by 2030. “This implies that the share of total US power demand accounted for by data centers will increase from around 3% currently to 80% by 2030,” relayed Davenport. It equates to “a 15% CAGR in data center power demand from 2023-2030.”
Chip Supply Limitations and Equity Forecasts
Let’s also not forget the problems of chip supply itself. The onset of the pandemic in 2020 rocked supply chains, creating significant shortages within semiconductor chips. It may be just a taste of the supply and demand imbalance these chips face due to ramping AI demand.
“We expect industry supply, rather than demand, to dictate AI chip shipments through 2H24 and into early 2025,” explained the Goldman Sachs U.S. semiconductor team. The team is led by Toshiya Hari, Senior US Semiconductor & Semiconductor Equipment Research Analyst. It includes Anmol Makkar and David Balaban, both semiconductor analysts. The team forecasts for “chip supply to eventually catch up with robust demand, though the next few years will likely prove painful amid the constraints in critical components.”
Christian Mueller-Glissman, Senior Multi-Asset Strategist at Goldman Sachs, took estimates one step further. Mueller-Glissman ran several models to forecast the impact of AI on equities. “Outside of the most bullish AI scenario that includes a material improvement to the structural growth/inflation mix and peak US corporate profitability, we forecast the S&P 500 returns would be below their post-1950 average.”
Image source: Goldman Sachs
What to Believe When It Comes to Generative AI
Overall, Goldman Sach’s report feels like an honest look at the path ahead for generative AI. There’s place for hype, but tempered enthusiasm seems, perhaps, the more realistic expression of that enthusiasm. There’s also ample place for skepticism but acknowledging the transformative changes happening holds weight as well.
At the end of the day, it’s up to each individual advisor and investor to determine their perspectives on generative AI’s disruptive potential. For now, Goldman Sachs appears to land on its relationship status with generative AI as a complicated one.
“We still see room for the AI theme to run, either because AI starts to deliver on its promise, or because bubbles take a long time to burst,” Goldman Sachs wrote. Not a resounding endorsement, but an honest one that I, for one, appreciate.
For more news, information, and analysis, visit VettaFi | ETF Trends.