Loomis Sayles & Co L P 13F Portfolio

The biggest add was Microsoft, where Loomis Sayles lifted the position by 28.3%, adding roughly $640.2M and pushing it to 3.64% of the book. That is a strong statement that the fund wants more exposure to the operating system of enterprise AI — cloud, productivity, and developer tooling — not just the GPUs that power it.

Broadcom tells the same story from the hardware side. The stake exploded by +1283.8%, adding about $305.3M and turning a rounding error into a meaningful 0.41% position, a clear attempt to own more of the networking and accelerator plumbing behind AI data centers.

They also leaned into application and workflow software: Autodesk (+10.4%), Salesforce (+6.5%), Workday (+10.4%) and a small increase in Shopify collectively deepen their bet that software vendors will be key monetizers of AI. Outside tech, adds to Novo Nordisk (+9.3%), Booking (+11.1%), Mastercard (+18.0%) and industrial names like Trane (+16.6%) and Cummins (+6.1%) suggest selective conviction in structural growth and pricing power, but in smaller dollar terms than the AI infrastructure push.

On the consumer side, a 5.4% add to Netflix and a sizable increase in MercadoLibre (+36.3%) show they still like scaled, asset-light platforms where incremental AI adoption can expand margins and engagement without heavy capex.

The trims are not a broad de-risking; they are a reallocation away from crowded or structurally challenged corners of the same themes. Nvidia was cut by 3.6%, unlocking about $295.2M from a position already up 3084.7% versus their average cost — classic profit-taking without changing the thesis that it remains a core winner.

The more aggressive cuts came in second-tier semiconductor and tools names. KLA was slashed by 38.3% (about $144.4M), and Taiwan Semiconductor was reduced by 25.7% (around $80.1M), effectively trading some exposure to upstream fab capacity and equipment for Microsoft and Broadcom.

In health care, Illumina stands out as a near-capitulation: they cut 34.4%, freeing roughly $177.8M from a position sitting 29.9% below their average cost. That looks less like disciplined trimming and more like a decision to stop averaging down in a structurally muddier story, especially when paired with modest reductions in Regeneron, Vertex, Novartis and Intuitive Surgical.

Elsewhere, small trims across Alphabet (both share classes), Visa, Monster Beverage, and a swath of financials (Goldman Sachs, JPMorgan, BlackRock, SEI) look chiefly like funding sources. These are long-held winners or stable franchises now partially monetized to back higher-conviction growth ideas.

On the surface, sector weights barely budged: technology crept from an estimated 44.85% to 45.39%, consumer discretionary from 16.76% to 16.79%, and industrials actually dipped slightly. But inside those buckets, the book rotated decisively toward the AI core stack and away from peripheral plays.

Within tech, they shifted from semiconductor breadth to a more curated architecture: more Microsoft and Broadcom, less Nvidia at the margin, and material reductions in KLA, TSMC and Arm. Software broadly benefited — Autodesk, Salesforce, Workday and Microsoft all grew — suggesting a view that recurring-revenue platforms will compound AI gains more steadily than cyclical chip capacity.

Health care shrank from 9.20% to 8.85%, with cuts in Illumina, Novartis, Regeneron and Intuitive partially offset by the Novo Nordisk add and a lift in Alnylam. That implies they see fewer near-term catalysts in tools and large-cap pharma versus GLP-1 and select high-science pipelines.

Consumer and financial exposure were tuned rather than reimagined. Consumer discretionary was reshuffled toward scalable, high-ROIC platforms (Netflix, Booking, MercadoLibre) and away from more cyclical or brand-sensitive names like Disney, Hilton and Nike. Financials slipped marginally as they trimmed banks and asset managers, even while adding to S&P Global, a higher-quality, data-and-indices compounder.

Put together, this is a portfolio saying that the next leg of equity returns will be driven by AI infrastructure and the software layers that sit directly on top of it. They are no longer just riding Nvidia and the Magnificent Seven; they are deliberately building a spine of hyperscale cloud, networking, and workflow software around that core.

At the same time, the incremental capital going into Novo Nordisk, Booking, Mastercard, MercadoLibre and industrial efficiency plays like Trane and Cummins shows they still believe in old-fashioned unit growth, pricing power and operating leverage. Those names provide a diversified set of earnings drivers that should hold up even if AI enthusiasm periodically overshoots.

The losers — Illumina, KLA, TSMC, and several financials — point to where Loomis Sayles has less patience: capital-intensive or structurally uncertain stories where returns are hostage to regulation, capex cycles, or single-product risk. They are willing to lock in losses or walk away from proven winners if that capital can be redeployed into what they see as the enduring infrastructure of the digital economy.

If the AI build-out and software monetization wave plays as expected, this quarter’s moves should give them a higher beta to that theme with a better quality mix. If not, their diversified consumer and industrial book may blunt the downside, but make no mistake: Loomis Sayles & CO L P is consciously steering this portfolio to be an AI-and-platforms growth engine rather than a balanced, benchmark-hugging core fund.