Blab AI
Bank of America Analyst Says AI-Driven Semiconductor Boom Is Still in Early Stages
On July 6, Bank of America Securities senior semiconductor analyst Vivek Arya told CNBC that the surge in demand for chips to power artificial intelligence (AI) is not a short‑term fad but a long‑term infrastructure buildout.
Arya said AI is moving from a proof‑of‑concept phase into a core part of enterprise software, and that the pace of investment in data‑center hardware is accelerating. The Philadelphia Semiconductor Index (SOX) has risen almost 80 % year‑to‑date through early July, a jump that reflects the rapid growth of the AI infrastructure sector. Historically, cumulative semiconductor sales reached $1 trillion after about 50 years; the next $1 trillion of sales was added in just five years, largely driven by the AI boom. The index’s performance underscores the scale of capital flowing into chip design, manufacturing and deployment.
Arya highlighted hyperscaler spending as the most visible indicator of demand. He noted that the combined capital expenditures of major cloud providers were estimated at $6–$7 billion at the start of the year, but are now approaching $1 trillion. If the trend continues, he projects next‑year capex could rise 40–50 % to $1.4–$1.5 trillion. The revenue base that supports this spending—annual recurring revenue from companies such as OpenAI and Anthropic—already exceeds that of some of the largest software firms, giving cloud operators a strong incentive to keep investing.
Company results further illustrate the shift. NVIDIA reported first‑quarter fiscal 2027 revenue of $81.6 billion, up 85 % from a year earlier, with data‑center sales of $75.2 billion, a 92 % increase. The company’s networking segment grew 199 % to $14.8 billion. CEO Jensen Huang described the expansion as “the largest infrastructure expansion in human history.” Supply commitments for the year total $119 billion, and Q2 revenue is guided to $91 billion.
Broadcom’s Q2 fiscal 2026 revenue reached $22.2 billion, up 48 %, with AI‑semiconductor sales of $10.8 billion, a 143 % rise. CEO Hock Tan guided Q3 AI‑semiconductor revenue to $16 billion, more than double the prior year, driven by custom accelerators and networking.
Taiwan Semiconductor Manufacturing (TSMC) reported May 2026 revenue of NT$416.98 billion, up 30 % year‑over‑year. The company’s shares are up 44 % year‑to‑date, and market participants estimate a 77 % probability that Q2 revenue will exceed $40 billion.
Advanced Micro Devices (AMD) saw its data‑center revenue grow from 14 % to 57 % year‑over‑year between Q2 2025 and Q1 2026, reaching $5.78 billion. CEO Lisa Su said customer engagement around the MI450 Series and Helios is strengthening, with forecasts exceeding initial expectations. Meta has committed up to 6 gigawatts of AMD Instinct GPUs.
Micron Technology posted Q3 fiscal 2026 revenue of $41.5 billion, up 346 %, with a gross margin of 84.6 %. CEO Sanjay Mehrotra said the results reflect the strategic value of memory in the AI era and cited multi‑year customer agreements. The company guided Q4 revenue to $50 billion.
Arya also noted that supply constraints are a sign of healthy demand. He said that chips deployed several years ago—such as Google’s TPUs, Amazon’s Trainium and NVIDIA’s GPUs—are fully utilized, leaving no idle compute capacity.
In summary, the semiconductor industry is experiencing a sustained, capital‑intensive expansion driven by AI workloads. The Philadelphia Semiconductor Index’s 80 % gain, rising hyperscaler capex, and strong quarterly results from NVIDIA, Broadcom, TSMC, AMD and Micron all point to a long‑term buildout rather than a short‑term hype cycle. While export restrictions and seasonal slowdowns may introduce volatility, the continued upward trajectory of cloud infrastructure spending remains the most reliable indicator of future demand.
Arya said AI is moving from a proof‑of‑concept phase into a core part of enterprise software, and that the pace of investment in data‑center hardware is accelerating. The Philadelphia Semiconductor Index (SOX) has risen almost 80 % year‑to‑date through early July, a jump that reflects the rapid growth of the AI infrastructure sector. Historically, cumulative semiconductor sales reached $1 trillion after about 50 years; the next $1 trillion of sales was added in just five years, largely driven by the AI boom. The index’s performance underscores the scale of capital flowing into chip design, manufacturing and deployment.
Arya highlighted hyperscaler spending as the most visible indicator of demand. He noted that the combined capital expenditures of major cloud providers were estimated at $6–$7 billion at the start of the year, but are now approaching $1 trillion. If the trend continues, he projects next‑year capex could rise 40–50 % to $1.4–$1.5 trillion. The revenue base that supports this spending—annual recurring revenue from companies such as OpenAI and Anthropic—already exceeds that of some of the largest software firms, giving cloud operators a strong incentive to keep investing.
Company results further illustrate the shift. NVIDIA reported first‑quarter fiscal 2027 revenue of $81.6 billion, up 85 % from a year earlier, with data‑center sales of $75.2 billion, a 92 % increase. The company’s networking segment grew 199 % to $14.8 billion. CEO Jensen Huang described the expansion as “the largest infrastructure expansion in human history.” Supply commitments for the year total $119 billion, and Q2 revenue is guided to $91 billion.
Broadcom’s Q2 fiscal 2026 revenue reached $22.2 billion, up 48 %, with AI‑semiconductor sales of $10.8 billion, a 143 % rise. CEO Hock Tan guided Q3 AI‑semiconductor revenue to $16 billion, more than double the prior year, driven by custom accelerators and networking.
Taiwan Semiconductor Manufacturing (TSMC) reported May 2026 revenue of NT$416.98 billion, up 30 % year‑over‑year. The company’s shares are up 44 % year‑to‑date, and market participants estimate a 77 % probability that Q2 revenue will exceed $40 billion.
Advanced Micro Devices (AMD) saw its data‑center revenue grow from 14 % to 57 % year‑over‑year between Q2 2025 and Q1 2026, reaching $5.78 billion. CEO Lisa Su said customer engagement around the MI450 Series and Helios is strengthening, with forecasts exceeding initial expectations. Meta has committed up to 6 gigawatts of AMD Instinct GPUs.
Micron Technology posted Q3 fiscal 2026 revenue of $41.5 billion, up 346 %, with a gross margin of 84.6 %. CEO Sanjay Mehrotra said the results reflect the strategic value of memory in the AI era and cited multi‑year customer agreements. The company guided Q4 revenue to $50 billion.
Arya also noted that supply constraints are a sign of healthy demand. He said that chips deployed several years ago—such as Google’s TPUs, Amazon’s Trainium and NVIDIA’s GPUs—are fully utilized, leaving no idle compute capacity.
In summary, the semiconductor industry is experiencing a sustained, capital‑intensive expansion driven by AI workloads. The Philadelphia Semiconductor Index’s 80 % gain, rising hyperscaler capex, and strong quarterly results from NVIDIA, Broadcom, TSMC, AMD and Micron all point to a long‑term buildout rather than a short‑term hype cycle. While export restrictions and seasonal slowdowns may introduce volatility, the continued upward trajectory of cloud infrastructure spending remains the most reliable indicator of future demand.
