a16z says AI's next growth cycle runs through chips, power and robots
Growth investor David George argues that AI demand is lifting chips, power and robotics while public software faces a tougher test on growth.
By Ryan Merket · Published
Primary source: Andreessen Horowitz
Why it matters
a16z ties its AI thesis to two tests founders can measure: whether compute demand keeps older hardware useful and whether software companies can pair profitability with renewed growth.
Andreessen Horowitz (a16z) published its second State of Markets report, arguing that technology's next growth cycle is moving beyond software and into the physical systems needed to build and run AI: semiconductors, power, networking, robotics and manufacturing. The report, published September 30th by a16z Growth leader David George, makes a case for where capital is moving, while also offering a venture investor's reading of the markets it hopes to fund.
George came to a16z in 2019 after seven years at General Atlantic, where he invested in companies including Airbnb, AppDynamics, CrowdStrike, Opendoor, Slack and Uber, according to a16z's biography. He now leads the firm's Growth practice. That background gives the report's software and public-market analysis a late-stage investor's emphasis: growth still earns a premium, but companies have to show that they can sustain it.
AI spending reaches the physical economy
The report's headline figure is its estimate that technology contributed about 76% of the S&P 500's total earnings growth in 2026 as of late August. a16z uses that figure to argue that tech has stopped behaving like one sector among many and has become a driver of the broader market cycle. The number is the firm's characterization; the article does not lay out the calculation or underlying dataset alongside the claim.
The next leg, in George's telling, is a shift from bits to atoms. AI data-center demand is flowing into chips, power and networking, while infrastructure expansion, defense spending, grid upgrades, manufacturing reshoring and robotics add other sources of demand. The report describes hyperscalers' cash flow as helping finance semiconductor capacity, with debt increasingly joining the mix.
That investment thesis fits a16z's own recent moves. On August 28th, the firm announced its $1.1B Machine Age Fund for AI's physical buildout. The report's case for hardware therefore does double duty: it describes a market trend and supports a direction in which a16z has committed capital.
The distinction between AI demand and mature AI adoption is visible in the report's own figures. a16z says nearly 30% of S&P 500 companies report some quantifiable AI impact, while only about 2% report a tracked metric. It also says that, as of April, roughly 2% of U.S. households paid for an AI service. Those estimates suggest substantial room for use to deepen, but they do not establish how quickly that growth will translate into returns for infrastructure investors.
Older GPUs still have a customer
George also challenges the argument that older processors will quickly become stranded assets as newer chips arrive. The report says rental rates for Nvidia A100 GPUs have held at or above their level at the start of the year, even as newer B200 systems draw strong demand. a16z's explanation is that cheaper AI compute encourages more use, allowing older chips to remain useful while total demand grows.
That is an important assumption for a market financing expensive hardware: the value of today's infrastructure depends partly on how much demand expands before the next generation arrives. The report presents rental pricing as evidence that older GPUs remain in use; it does not establish that every chip, operator or data center will retain its value. a16z itself says the story is still unfolding.
Public software gets a growth test
The report is more measured on software than the loudest predictions of AI-driven displacement. It says roughly 75% of public software companies are profitable, but only about 30% are growing at 20% or more. a16z's explanation for the sector's repricing is that companies shifted toward profitability as capital became more expensive, leaving fewer businesses with the growth rates that historically supported higher multiples.
That framing makes the report less a forecast of software's disappearance than an argument about selection. Fast-growing software companies, George writes, still trade at multiples closer to historical averages, while slower growers have faced a broader reset. For founders, the practical pressure is familiar: profitability can help a company endure, but it does not automatically restore the valuation that came with faster growth.
The first State of Markets report appeared in January and focused on public and private markets and AI. The second edition extends that view through the first two quarters of 2026 and puts the hardware buildout, lingering GPU demand and software's growth slowdown in the same frame.
George's closing forecast is that AI will broaden demand across enterprise and consumer markets and into robotics, biotech and health. That is a wide field for a growth investor to survey. The report's more concrete argument is narrower: the next phase of AI investment will depend on whether demand reaches the physical infrastructure behind the models, and whether the companies building that infrastructure can turn a capital-heavy expansion into durable earnings.