State Board Of Administration Of Florida Retirement System 13F Portfolio

The most aggressive add is in KLA, where the fund increased shares by 875.0%, lifting the position to 0.56% of the book and adding about $304.1M of exposure. That is a blunt statement that process control and yield management – not just GPUs – are where incremental AI dollars show up in earnings over the next few years.

They also pushed further into SanDisk and Marvell, both deep in the memory and connectivity stack that AI data centers are starving for. SanDisk is up 3.1% in shares and now sits around $264.6M, while Marvell, already a big AI networking and custom silicon story, was nudged higher again.

Palo Alto Networks got another 2.7% in shares, adding roughly $6.1M and underlining that securing AI- and cloud-heavy enterprises is now as core to the thesis as compute capacity itself. This isn’t a speculative swing: the position is already up 551.7% versus their average cost, and they’re still adding.

The other notable add is Tesla, up 1.7% in shares and about $17.4M in value, framed here less as a carmaker and more as an industrial automation and AI platform. Taken together, the “biggest buys” table reads like a checklist of bottlenecks to scaling AI: lithography yield, memory bandwidth, network fabric, cybersecurity, and applied robotics.

Funding for this push clearly comes from trimming the “AI royalty” tier: Nvidia, Apple, Microsoft and both Alphabet share classes are all down low-single digits in share count. Nvidia alone freed up about $138.7M despite still commanding 6.6% of the book and showing a 4,257.0% gain versus their average cost.

The pattern is consistent across the rest of the Big Tech complex. Apple, Microsoft, Alphabet and Amazon all saw modest reductions in shares (roughly -1.9% to -2.8%), monetizing enormous embedded gains while leaving the core growth engines firmly in place.

Second-tier semis like Micron, AMD, Intel and Applied Materials were also lightly shaved rather than abandoned, suggesting the manager sees more targeted upside in tools (KLA, Lam) and specialized connectivity (Marvell) than in broad beta to the chip cycle. Even Eli Lilly – a poster child for GLP-1 euphoria with a 1,510.7% gain vs cost – was gently trimmed, treating it as another profit pool to redirect toward under-owned AI infrastructure.

Traditional cyclicals and financials show a similar “use as ATM” profile: JPMorgan, Bank of America, Goldman Sachs, Exxon and Chevron all appear in the biggest trims list, with mid-single digit share cuts that loosen capital from lower-growth, more macro-sensitive exposures.

The sector chart makes the philosophy obvious: this is a growth-heavy, AI-centric equity book whose risk budget lives in technology. Tech edges up to 62.01% of the portfolio even as the manager takes money off the top of Nvidia, Apple, Microsoft and Alphabet.

Consumer discretionary, at 10.43%, barely moves, but the internal mix matters: Amazon, Costco, Home Depot and Netflix are being slightly trimmed, while Tesla – classified here as industrials – is the consumer-adjacent name that’s actually growing. It’s a tilt away from pure consumption and toward platforms that produce AI-enabled goods and services.

Health care, finance, energy and staples all leak a few basis points of weight, with banks, big pharma, integrated oils and Coca-Cola consistently on the trim list. Real estate exposure is effectively a mislabel for Visa and Mastercard, both of which were cut by around 3–5% in shares, quietly reducing payment cyclicality.

Industrials creep up to 4.91% from 4.85% on the back of Tesla, even as Caterpillar and RTX are trimmed. The message from the bar chart: the fund is comfortable letting its ballast sectors gradually shrink while concentrating incremental dollars in the infrastructure and security layers of the AI economy.

Taken together, this quarter’s moves sketch a clear roadmap: stay long the AI platform leaders, but push the marginal dollar into the supply chain choke points and real-economy adopters. Trimming Nvidia and friends by low-single digits while lifting KLA almost tenfold is a bet that returns will diffuse down the stack as capex broadens beyond a handful of GPU vendors.

The modest but deliberate increases in SanDisk, Marvell and Palo Alto suggest they see data intensity, network complexity and security as secular, not cyclical, features of the next leg of the market. Tesla’s incremental add fits the same story: AI is leaving the data center and colliding with manufacturing, logistics and consumer hardware.

On the risk side, the slow bleed from banks, energy, staples and even traditional health care implies less reliance on value and yield as hedges. With tech above 60% and the top 10 at 32.9% of the book, this remains a conviction-heavy expression of the AI and cloud buildout.

If that thesis is right, expect future 13Fs to keep rotating within tech – away from crowded, fully rerated winners and deeper into tools, memory, networking and security – rather than meaningfully rebuilding exposure to classic defensives. If it’s wrong, this is a fund that has chosen to live and die by the AI infrastructure cycle.