Sei Investments Company 13F Portfolio

If you want to know what Sei still believes in, follow the add list: they are not backing away from AI; they are just being choosier about how they own it.

• NVIDIA (3.59% of the book) was increased by 4.1%, a roughly $159.5M add despite already sitting more than +256.0% above their average cost. That is not risk reduction — it’s an explicit statement that, in their view, the AI leader’s earnings power still outruns the hype.
• Micron, up more than +409.2% versus their average buy, was also increased by 7.1% (about $65.7M). This leans into the idea that AI memory demand and HBM scarcity have durable legs, even after a dramatic re-rating.
• On the “airbag” side, the biggest dollar adds are BND (+7.8%, about $106.5M), VTEB (+11.8%, about $84.9M) and BNDX (+7.1%, about $42.3M) — broad taxable, muni, and international bond exposure. All three sit slightly underwater versus cost, so Sei is averaging into duration rather than chasing past returns.
• NXUS (+10.7%, about $69.7M) and IDEV (+4.4%, about $48.7M) push more capital into international equities via ETFs, while ACWV (+3.4%) and SEIQ (+5.3%) add a low-volatility and multi-asset tilt.

Taken together, Sei is saying two things at once: the AI infrastructure cycle still has room, led by the highest-quality chip names, and the right complement is not more beta, but carry and diversification in bonds and global ETFs.

The sell tape is just as revealing: this is classic gain-harvesting in the frothiest corners of tech to pay for the bond and ETF build-out.

• Among the megacaps, Apple was cut by -24.9% (about -$811.1M), Microsoft by -15.5% (about -$405.8M), Amazon by -10.0% (about -$233.0M), and Alphabet’s two share classes by smaller single-digit trims. All four still show large gains versus cost, so this looks like profit-taking and concentration management, not a structural rejection of US big tech.
• In semis and semi-cap, the trims are sharper: AMD (-26.6%, roughly -$269.5M), TSMC (-18.0%, about -$223.0M), Broadcom (-12.1%, about -$231.5M), and Applied Materials (-22.8%, about -$108.0M) all saw meaningful cuts. Lam Research and ASML were also reduced. Sei appears to be rotating within the AI value chain, keeping or adding to the names where they think upside is still mispriced (NVIDIA, Micron) while dialing back anything they see as fully valued infrastructure.
• Outside of tech, they lightened up on payments (Mastercard, -22.1% or about -$140.9M), autos (Tesla, -10.9%), pharma (Eli Lilly, Johnson & Johnson), and JPMorgan. These are classic funding sources: profitable, liquid winners that have done their job.
• Even their own SEI-branded ETFs (SEIM, SEIV) were modestly reduced, while cheaper, broader vehicles like Vanguard and Schwab funds got incremental capital.

This is not a panic de-risking; it’s surgical. Sei is taking chips off the table where multiples and positioning look crowded, and redeploying into duration, munis, and the AI names they think still have runway.

Under the hood, the sector mix is shifting from a tech-heavy growth posture toward a barbell of AI plus diversified ballast.

Technology still carries the book at 42.27%, but that’s down from an estimated 44.64% as they trimmed a wide swath of megacap and semi-cap names. Consumer discretionary, health care, real estate, finance, and industrials all edge down slightly, mostly via trims in Amazon, big pharma, the card networks and Tesla.

The winner is the so-called “unclassified” bucket, which in reality is a stack of broad ETFs and bond funds. That sleeve rises to 42.95% from 40.03%, powered by adds to BND, VTEB, BNDX, SPHY, SCHO, and a range of Vanguard, Schwab, iShares and Nuveen equity ETFs (VEA, SCHG, SCHV, VTV, VUG, IEMG, VWO, IEFA, ACWV and others).

Conceptually, Sei is dialing down idiosyncratic equity risk and leaning more on index-like, rules-based exposure. They are not abandoning growth — SCHG, VUG and the continued NVIDIA/Micron concentration prove that — but they are migrating part of the book into structures that can better weather single-name blow-ups or an AI sentiment swing.

For an allocator looking at this book, the message is that Sei wants the AI upside, but within a more benchmark-shaped, less fragile chassis.

Put the quarter together and the forward thesis is clear: Sei expects the AI cycle to continue driving equity returns, but is no longer willing to run it at maximum throttle.

Keeping NVIDIA as the top position and adding to Micron while trimming AMD, TSMC, Broadcom and Applied Materials says they are discriminating inside the AI complex rather than de-grossing it. The portfolio is being rewired around a smaller set of high-conviction beneficiaries, against a backdrop of sizable unrealized gains.

At the same time, the stepped-up buying of BND, VTEB, BNDX, SPHY and short-duration ETFs like SCHO shows they want more predictable carry and ballast if equity multiples compress or the rate path surprises. Additional flows into international and low-volatility equity ETFs hint that they see better risk-adjusted return outside the narrow US megacap growth cohort they just harvested.

Going forward, expect Sei to keep running this AI-and-airbag barbell: a concentrated core in best-of-breed compute and memory, framed by a growing ring of bonds and diversified ETFs. If volatility spikes or AI leadership broadens, this setup gives them room to either reload selectively into weakness or simply let the cushion do its job while the longer-term AI earnings story plays out.