David Solomon has heard the warnings. Mass unemployment. White-collar wipeout. A future where machines handle the thinking and humans scramble for scraps. The Goldman Sachs chief executive calls those fears overblown.
In a New York Times guest essay published May 22, Solomon framed artificial intelligence as the latest chapter in a long story of technological change. Electricity, computers, spreadsheets. Each displaced tasks. Each created more value elsewhere. AI follows the same path, he argued. It does not end work. It redirects it.
“Fears of a job apocalypse may very well overlook A.I.’s potential to spur an economic and productivity revival,” Solomon wrote. Short. Direct. The kind of line that lands with investors scanning for conviction.
Goldman’s own research backs the view. The bank estimates AI could automate roughly 25 percent of current work hours over the next decade. That sounds large until you consider the three forces at play. Routine tasks disappear. Performance standards rise across remaining roles. And entirely new positions emerge to manage, audit and govern the systems themselves. The result looks less like contraction and more like reallocation.
Entry-level jobs already feel the squeeze. Financial modeling, note-taking, basic spreadsheet work. Junior analysts watch tools handle what once filled their days. But the bank sees demand shifting toward relationship-driven work in investment banking, trading and asset management. Data-center construction has already supported hundreds of thousands of U.S. jobs since 2022. Creative destruction, Solomon calls it. Not collapse.
If disruption hits harder than expected, he suggests a joint effort between public and private sectors to help workers adapt. No panic. Just preparation.
Yet the infrastructure side tells a different story. One measured in trillions.
Goldman Sachs Research laid out the numbers in early May. A baseline case points to $7.6 trillion in cumulative capital spending on AI from 2026 through 2031. That covers specialized chips, data centers and power generation. Annual outlays start near $765 billion in 2026 and climb to $1.6 trillion by 2031. The report stresses these figures rest on supply-side assumptions that can shift dramatically.
Chip useful life matters. So does the cost of next-generation data centers. Architectural choices between GPUs and specialized processors can swing silicon budgets. Physical bottlenecks around power, labor and equipment could stretch timelines without reducing total spend. Innovation remains the biggest unknown. A breakthrough that cuts compute needs for training or inference would rewrite the entire equation.
Hyperscalers already signal the scale. The largest cloud operators plan to pour more than $500 billion into capital expenditures in 2026 alone. The seven biggest technology companies now represent more than 30 percent of the S&P 500’s market capitalization and about one-quarter of its earnings. Their spending drives much of the momentum.
Power stands out as the binding constraint. Goldman Sachs Research projects data-center electricity consumption will jump 175 percent by 2030 from 2023 levels. A separate commodities update forecasts U.S. data-center power demand will more than double to 66 gigawatts by 2027 from 31 gigawatts in 2025. That analysis highlights how new construction must meet accelerating AI workloads even as grid connections lag.
These forecasts have evolved. Earlier projections called for a 165 percent rise in global data-center power demand by 2030. The latest base case pushes higher. Grid access, interconnection queues and on-site generation now decide who scales fastest. Renewables help on the margin but struggle to deliver the 24/7 reliability data centers demand without major battery advances.
Solomon himself has struck bullish notes on the broader economy. In February he described an AI supercycle paired with deregulation as powerful tailwinds for growth and dealmaking. He predicted unprecedented IPOs in 2026 fueled by private-equity dry powder and technology investment needs. The comments reflected confidence that capital formation around AI would lift activity across markets.
Inside Goldman the bank puts its money where its analysis points. The firm spent $6 billion on technology in 2025 and signaled appetite for more. Chief Financial Officer Denis Coleman said in April that early results from an internal overhaul called One Goldman Sachs 3.0 reinforced the need to accelerate cloud migration and improve data quality. Those steps aim to unlock AI deployment across six initial work streams. The earnings commentary showed net income rising 19 percent year-over-year even as the bank invested heavily.
Solomon has also spoken about headcount. AI does not mean fewer people. It means the bank can afford more high-value talent. Productive workers become even more productive. The enterprise grows larger.
Yet caution appears in places. Last year Solomon warned of possible market drawdowns if capital deployed into AI fails to generate expected returns. He drew parallels to past manias. More recently he described a sharp sell-off in software stocks as too broad, arguing winners and losers would emerge but many companies would pivot successfully.
The bank’s latest research on corporate concentration adds another layer. History shows technological change tends to widen gaps between leading firms and the rest. Scale and network effects reward those who invest successfully in intangibles. AI looks likely to follow that pattern. Goldman economists reached that conclusion after reviewing nearly a century of income, sales and tax data.
Recent moves show Goldman participating directly in the buildout. In mid-May the bank joined BBVA and other investors in backing DeployCo, an OpenAI venture aimed at helping large organizations embed AI into core operations. The deal exceeded $4 billion and reflects confidence in enterprise adoption. American Banker reported that nearly three-quarters of banks increased AI investment last year.
Looking ahead, Goldman’s own CIO has pointed to 2026 as a breakout year for personal AI agents. These systems could handle routine tasks from rebooking flights to managing schedules. The bank expects agent-as-a-service models to emerge with token-based economics. Memory and context will matter more than raw model size. Learning becomes the essential human skill.
Power limits could impose a gigawatt ceiling on rapid expansion. Alliances among technology leaders may harden into duopolies as network effects compound. The infrastructure race favors those who secure energy, land and supply chains early.
Solomon’s message remains measured. Excitement around AI is justified by its potential impact. The spending surge looks real. Returns will vary. Some capital will prove unproductive. Markets may recalibrate. Over the long run the net benefits should accrue to institutions that invest wisely.
The contrast stands out. On one side, dire predictions of jobless futures. On the other, a bank forecasting trillions in spending, rising productivity and adaptation through new roles. Goldman’s analysis does not dismiss disruption. It simply refuses to treat it as apocalypse.
That stance carries weight coming from an institution that advises the companies making these bets. It also shapes how clients think about workforce strategy, capital allocation and competitive positioning. The data-center cranes keep rising. The models keep improving. And the conversation keeps circling back to the same question. How much changes, and how quickly can the economy absorb it?
So far Solomon bets on absorption. History, he says, supports that bet. The numbers Goldman publishes suggest the test will be enormous.
WebProNews is an iEntry Publication