The Harbor AI Inflection Strategy ETF (filed December 8, 2025) is an actively managed, non-diversified fund seeking long-term total return. It carries a management fee of 0.88% with no 12b-1 fees, putting total expenses at 88 basis points. That's not cheap for an ETF, but it's competitive for an actively managed thematic strategy with a boutique subadvisor. Harbor's 'unified fee' structure means the advisor covers most operating expenses, so what you see is largely what you get.
The fund will hold 30-50 positions and concentrates in information technology and industrials (25%+ in each). It can invest across all market caps, with up to 20% in non-U.S. companies including emerging markets.
Here's where it gets interesting. The prospectus doesn't just wave hands about 'AI companies.' It explicitly defines four categories of holdings:
Infrastructure Enablers: Data centre real estate and construction, power generation and cooling systems, computing power, networking and storage. This is the picks-and-shovels layer that's been minting money for companies like Vertiv, Eaton, and the data centre REITs.
Middleware Platforms: Cloud service providers, AI platforms and tools, and enterprise software integrators. Think the Microsofts, Snowflakes, and Palantirs of the world.
Adopters: Companies using AI to enhance operations, reduce costs, or create new revenue streams across healthcare, financial services, industrial and manufacturing, tech, and consumer/retail. This is the 'inference economy' play that everyone keeps talking about.
Second-Order Beneficiaries: Energy and utility companies, professional services, training and consulting firms, and cybersecurity providers. The 'wait, they benefit from AI too?' category.
The prospectus explicitly states the portfolio mix will shift as 'new technologies supporting AI growth emerge (e.g., novel chip architecture, quantum hardware or next-generation cooling solutions).' So 'inflection' apparently means they'll actively rotate between these layers as the AI buildout evolves. That's an actual strategy, not just marketing copy.
The subadvisor isn't named in the summary prospectus, but the investment philosophy is unmistakably EARNEST Partners' fingerprint. The strategy description matches almost verbatim with Harbor's other EARNEST-managed funds: bottom-up fundamental analysis, emphasis on 'strong business fundamentals' (revenue growth, profit margins, manageable debt, strong cash flow), and a focus on 'earnings prospects that are not recognized by the market.'
EARNEST Partners is an Atlanta-based, 100% employee-owned firm managing over $50 billion. They're not a quant shop throwing AI buzzwords at a screen. Their approach is decidedly old-school: find companies with good fundamentals that the market is underpricing, hold conviction positions, and sell when either the thesis plays out or something better comes along.
The sell discipline is explicit: exit when the stock reaches the subadvisor's price target, when a better opportunity emerges, or when the company's prospects deteriorate. This is value-with-a-growth-overlay investing applied to AI. Not exactly what you'd expect from something called 'Inflection Strategy.'
The prospectus includes a dedicated 'Artificial Intelligence Risk' section that's refreshingly honest. Some highlights: 'Companies seeking to enable and/or adopt AI may fail to do so successfully.' 'Issuers engaged in AI typically have high research and capital expenditures and, as a result, their profitability can vary widely, if they are profitable at all.' 'It can be difficult to accurately capture what qualifies as an AI company.'
That last admission is almost endearing. Even the fund manager is acknowledging that 'AI company' is a fuzzy concept. They also note regulatory risk, intellectual property concerns, and the fact that many holdings will have business lines unrelated to AI that could affect performance. This is a fund that knows what it doesn't know.
The bull case for this fund isn't about catching the next Nvidia. It's about professional-grade rotation across the AI value chain as different segments mature. Most AI ETFs are effectively semiconductor funds with some cloud exposure bolted on. They're riding the infrastructure buildout phase and will likely suffer when that spending decelerates.
By explicitly including 'second-order beneficiaries' and 'adopters,' this fund has the mandate to shift toward companies monetizing AI rather than just building it. That's the transition everyone says is coming, but few AI funds are structured to capture.
The concentrated portfolio (30-50 names) and non-diversified structure mean the managers can take real positions rather than index-hugging. Combined with EARNEST's value discipline, you could end up with an AI fund that actually cares about valuations. Novel concept.
The obvious risk: timing these rotations is hard. Really hard. If the infrastructure buildout continues longer than expected, a fund positioned for the 'adoption' phase could underperform SMH for years. Markets have a nasty habit of staying irrational longer than active managers can stay solvent.
At 88 basis points, you're paying meaningful fees for this active management. If the manager's layer-rotation calls are wrong, you're underperforming cheaper alternatives while paying for the privilege. The concentrated portfolio cuts both ways: conviction positions generate alpha when you're right and significant underperformance when you're wrong.
There's also the 'Unrelated Business Risk' they disclose: many portfolio companies have business lines outside AI that could drag on performance. Your 'AI adopter' in healthcare might get hammered by a drug trial failure that has nothing to do with their AI initiatives.
The Harbor AI Inflection Strategy ETF is a more thoughtful product than its buzzword-laden name suggests. It's not trying to be another SMH clone or a pure-play AI hype vehicle. It's an actively managed, value-conscious strategy designed to rotate across AI's value chain as the market evolves.
Whether that rotation will be timed correctly is the $65 billion question (Harbor's current AUM). But at least they're trying something differentiated. In a market drowning in AI ETFs that all hold the same seven stocks, that counts for something.
It'll be interesting to see how the initial portfolio is positioned: heavy on infrastructure enablers riding the capex wave, or already tilting toward adoption plays betting on the next phase?
Either way, I owe Harbor a partial apology. The name is still peak 2025 buzzword syndrome, but the strategy underneath is genuinely interesting.