According to Bank of America analyst Vivek Arya, the artificial intelligence boom is growing rather than slowing down. Arya stated that the sector is only at the "midpoint" of a ten-year change, which is being spearheaded by Nvidia (NVDA) and Broadcom (AVGO), despite the fact that AI doubters have used eye-popping values as an excuse to flee.
Arya predicted a 30% year-over-year increase in global semiconductor sales in a report titled "2026 Year Ahead: choppy, still cheerful," which will ultimately propel the industry above a historic $1 trillion annual sales milestone in 2026.
Arya saw a high preference for businesses that had "moats that are quantified by their margin structure." He named four more large-cap semiconductor companies as his top choices for 2026, in addition to Nvidia and Broadcom: Lam Research (LRCX), KLA (KLAC), Analog Devices (ADI), and Cadence Design Systems (CDNS).
"I often say that investing in semis is very simple," Arya said to reporters during a call on December 19. "A sell-side analyst is not required to accomplish that.
You won't be all that incorrect if you take all of your businesses, organize them by gross margins, and purchase the top five."
According to BofA, the addressable market for AI data center systems is expected to increase at a compound annual growth rate of 38% and reach over $1.2 trillion by 2030. The potential for AI accelerators alone is $900 billion.
The market is still wary despite these startling numbers because AI data centers are costly. According to Bank of America, a typical 1 gigawatt facility needs capital expenditures of up to $60 billion, of which about half goes straight to hardware.
Arya maintained her optimism, claiming that present expenditures are both "offensive and defensive." Put differently, Big Tech is forced to make investments in order to safeguard its current empires.
The largest firm in the world by market capitalization, Nvidia, is presently functioning in a "different galaxy," according to Arya. Arya cautioned against equating the AI leader with conventional chipmakers, despite Nvidia's stock rising more than 40% so far this year.
An Nvidia graphics processing unit (GPU) costs about $30,000, but the average chip costs $2.40. Additionally, BofA cited free cash flow, which is expected to reach half a trillion dollars over the next three years, and a valuation that is "still incredibly cheap" when growth is taken into account, despite some concerns that Nvidia's market capitalization has reached a ceiling.
In comparison to the larger S&P 500 (^GSPC), which trades at about two times its price-to-earnings growth (PEG) ratio, Nvidia appears to be a good deal at about 0.6x.
Broadcom is the nerve system if Nvidia is the AI brain. With a $1.6 trillion market capitalization, Broadcom has transformed from a component supplier to a pillar of AI infrastructure, with shares up more than 50% so far this year.
Its unique application-specific integrated circuits (ASICs) for hyperscalers like Google (GOOGL, GOOG) and Meta (META) are responsible for the change. These behemoths are looking to Broadcom in an effort to lessen their need on Nvidia.
The sentiment is shared by others on Wall Street. James Schneider, an analyst at Goldman Sachs, saw Broadcom as a vital arms dealer in the AI boom in a research report.
Schneider emphasized the company's exceptional capacity to control bespoke silicon with a price goal of $450, pointing to additional "upside" from growing partnerships with AI firms like Anthropic (ANTH.PVT) and OpenAI (OPAI.PVT).
Arya acknowledged that despite the excitement, the path to $1 trillion will be "choppy" and that no stock is "riskless." His top six for 2026, however, were picked especially because of their commanding market shares, which usually range from 70% to 75%.
"Look at the leaders in any part of technology, and you will usually see the leader has that kind of market share," Arya said. "It's actually the norm."
