Massachusetts Financial Services Company 13F Portfolio

The biggest adds table reads like a shopping list for the AI build‑out rather than a momentum chase in the obvious poster children. KLA, Micron, Broadcom and Arista are the clearest tells that MFS wants exposure to the plumbing of the data and AI cycle.

• KLA (1.42%, up +793.5% in shares, +$3.97B) is the statement trade. They turned a modest position into a top‑tier AI capital‑equipment bet even though the stake sits at a -48.0% mark‑to‑cost. That’s not averaging up into strength; it’s a willingness to increase size in a structurally critical, temporarily out‑of‑favor name.
• Micron (+2109.3% in shares, +$1.51B) adds a major memory and storage leg to the thesis. With a +20.9% gain vs average cost they’re buying into strength, signalling they see AI‑driven DRAM/HBM demand as early innings, not late.
• Broadcom (+34.6% in shares, +$1.40B) and Arista (+34.6%, +$439.8M) extend the theme from compute to connectivity. Those adds say they expect AI traffic and custom ASIC demand to keep compounding even if headline GPU names cool.
• Alphabet (+22.6%, +$1.94B) and Amazon (+23.6%, +$1.69B) are the platform complements: cloud, search and e‑commerce scale that monetize AI in production. Notably, Amazon is still -85.9% vs MFS’s average cost, so they’re doubling down on business momentum, not paper gains.
• Smaller but telling, CAT (+10.4%, +$177.4M) and LIN (+3.5%, +$109.2M) give them exposure to industrial and process bottlenecks that also benefit from capex and reshoring tied to data centers and manufacturing.

Pull these together and the pattern is consistent: more capital into enablers of data center, semiconductor and networking build‑out, and into cloud platforms that can actually turn that infrastructure into revenue growth.

On the sell side, the message is fund the next leg of AI from yesterday’s darlings and over‑owned defensives. The biggest trims by dollars are Microsoft, Progressive, JPMorgan, Johnson & Johnson and, notably, NVIDIA itself.

• Microsoft (-19.8%, -$2.16B) is still a 2.78% position with a massive +365.1% gain vs cost. MFS is clipping gains in the consensus AI platform winner to recycle into semis and tools where the risk/reward looks less fully priced.
• NVIDIA (-4.2%, -$605.2M) is similar: at 4.33% it remains the single largest line item, but the modest trim after a +367.2% gain vs average cost says they now see it more as a cash source than a place to increase risk.
• Progressive (-36.7%, -$963.8M), JPMorgan (-16.3%, -$784.6M), Morgan Stanley (-21.6%, -$528.4M) and Nasdaq (-20.0%, -$457.4M) show them reducing exposure to rate‑sensitive and market‑beta financials after a strong run. With JPM showing a +584.1% gain vs cost, this is classic profit‑taking.
• Meta (-18.3%, -$462.8M) and Amphenol (-24.7%, -$632.2M) fit the same pattern in tech: high‑quality, but crowded names where AI upside may already be embedded. Trimming Meta while adding Alphabet suggests a preference for search and cloud economics over social advertising at this point in the cycle.
• In Health Care, they’re quietly shrinking J&J (-21.6%, -$742.6M), Humana (-20.2%, -$420.6M) and MCK (-15.6%, -$344.4M), all with solid positive P&L, freeing up capital from stable earners to redeploy into higher‑beta growth.

Across these sells, there’s no panic, just a deliberate redistribution: less in fully rerated compounders and balance‑sheet havens, more in the upstream capital‑spending beneficiaries of AI and data growth.

The sector chart confirms what the individual trades suggest: MFS is structurally overweighting technology at the expense of financials and classic defensives. Technology climbs to 45.28% of the portfolio from 42.03%, an outsized move for a firm this large.

Finance drops to 14.83% from 17.13% as they lighten JPMorgan, Morgan Stanley, Schwab, Nasdaq, American Express, Chubb and Progressive. This is a broad reduction across money‑center banks, brokers, card networks and insurers rather than a single‑name call.

Health Care ticks down to 10.10% from 11.37% even though it remains a core ballast, with cuts across J&J, Medtronic, Cigna, Humana, BDX and McKesson. Real Estate (including card networks mislabeled as Real Estate in the data) edges down to 4.42% from 4.85%, helped by reductions in Visa, Mastercard and Prologis.

Energy slips to 2.60% from 2.89% on trims to Exxon and ConocoPhillips, and Utilities ease to 2.14% from 2.33% via Duke and PG&E. Meanwhile, Consumer Discretionary grows to 6.68% from 5.54%, almost entirely via the larger Amazon stake.

Even within tech, the mix is changing: they’re cutting Microsoft, Meta, TSMC, Analog Devices, NXP and Amphenol, while adding KLA, Micron, Broadcom, Arista, Alphabet and modestly Eaton and Seagate. The tilt is away from mature or fully priced cyclicals and toward capital equipment, memory and cloud‑centric names most leveraged to the next wave of AI spending.

Look through the noise of individual names and the thesis is clear: MFS is betting that AI‑driven capex, data intensity and cloud monetization will outgrow financials and bond‑proxies over the next few years. The decisions this quarter consistently favor cyclically sensitive tech over interest‑rate beneficiaries.

By turning KLA, Micron, Broadcom and Arista into much more substantial lines, they’re positioning for a multi‑year equipment and networking cycle rather than a one‑off GPU boom. The willingness to add aggressively to KLA while it’s underwater underscores a time horizon measured in years of wafer‑fab and inspection demand, not quarters of earnings beats.

At the same time, trimming Microsoft, NVIDIA and Meta — but keeping them large — shows a view that AI alpha is shifting from obvious leaders to second‑derivative beneficiaries, not that the AI theme is over. Alphabet and Amazon adds reinforce that the monetization layer in search, cloud and commerce still has room to compound.

The broad de‑weighting of banks, brokers, card networks, utilities and defensive health care reads as a macro call: less confidence that elevated rates and spread income can keep driving financial stocks, and less appetite for “safe” equities now that growth has re‑asserted itself. If this quarter is a guide, future moves will likely keep rotating capital within tech — from winners that have already re‑rated to the less‑owned enablers of compute, storage and connectivity — while keeping cyclicals and defensives as risk management tools, not return engines.

For observers, the key takeaway isn’t that MFS is a tech bull — that’s old news. It’s that they are explicitly upgrading their AI bet from headline stories to the harder‑to‑own infrastructure names that have to get paid if this cycle is real.