The alpha Brief: Insights on Quant Investing and the alpha Brief Portfolio (aBP)

The most important AI story this week was not a new benchmark or a single chip launch. It was a consumer product. Meta's Muse gave investors a tangible example of what mass market AI agents could look like, and that changed the conversation around the companies that distribute and supply the required compute.

The first phase of the artificial-intelligence trade was about securing enough compute. The second was about building the networks, data centres and power systems required to use that compute at scale. The next constraint may be smaller, less visible and just as important.

I am downgrading Penguin Solutions from a STRONG BUY to a HOLD. Execution improved, but the rerating from $32.40 to $53.30 absorbed most of the near-term upside. Q3 FY26 net sales reached a record $479Mn and adjusted EPS was $0.84, beating consensus by 13.6% and 50.0%, respectively.

The operating thesis strengthened. Production reached 4.07 Mn boe/d, the company achieved its $3 Bn cost-saving run rate early, and Hess synergies were raised to $1.5 Bn.

I am maintaining a BUY rating on Movado, but conviction is lower. At roughly 16.7x [my FY2027 EPS estimate], the shares no longer qualify as deeply discounted.Q2 FY2027 net sales rose 4.9% to $169.8Mn and adjusted EPS reached $0.54. Excluding the $0.11 duty-refund benefit, adjusted EPS was still about $0.43, up sharply from $0.23 a year earlier.

Power generation, grid equipment and critical minerals are becoming increasingly interconnected. Reliable electricity cannot be delivered without generation capacity, transmission infrastructure and the raw materials required to build both.

My US$94 present target uses 8.0x 2027E EBITDA and a 10% discount rate. At 9.3% price upside, the risk-adjusted return fits a Hold rather than a Buy.

I am downgrading Astec from a BUY to HOLD. Astec Industries reported Q2 2026 sales of $408.1Mn, up 23.6%, and adjusted EBITDA of $42.6Mn, up 26.0%. Adjusted EPS of $0.94 missed the $1.04 consensus estimate.

I downgrade Coherent from Strong Buy to Buy. The business case is stronger, but 15.5% expected upside and poor cash conversion no longer justify the top rating.

Q2 revenue rose 21.1% to $391.1 Mn and adjusted EBITDA rose 21.7% to a record $216.8 Mn. Revenue beat consensus, but adjusted EPS of $0.55 missed the $0.67 estimate.

I am downgrading Dana's rating from a STRONG BUY to a BUY. For Q2 Dana delivered sales of $2.01Bn, adjusted EBITDA of $207Mn and a 10.3% adjusted EBITDA margin. Revenue rose 4% and the margin expanded 270 basis points y/y.

I am downgrading KARO from BUY to HOLD. It's execution is better, but I believe that the former valuation discount has largely disappeared and Q1 cash conversion needs watching.

I downgrade GE Vernova from Buy to Hold because stronger execution is now balanced by a demanding valuation and unusually working-capital-heavy cash flow.

Oracle, Ciena and Credo sit at different points in the AI capital cycle. The investment question is shifting from who can build the infrastructure to who can earn an attractive return from it.

I am downgrading SSRM from a STRONG BUY to BUY. The balance sheet and capital returns strengthened. But the operating margin of safety weakened because costs and second-half execution now carry more weight.

I maintain a Strong Buy on Credo Technology Group Holding Ltd since it has reported fiscal Q1 2027 revenue of $479.0Mn, up 9.6% sequentially and 114.7% Y/Y. Non-GAAP EPS of $1.20 beat the $1.17 consensus by $0.03.

Our three stocks to watch this week capture different stages of the AI value chain. Vertiv addresses one of AI's increasingly important constraints which is power availability while Lam Research supplies the equipment needed to manufacture advanced semiconductors and Microsoft provides the cloud and software layer through which AI can ultimately be monetized.

AI's latest rally wasn't driven by a single stock. It spread across three critical bottlenecks that power the entire AI ecosystem: connectivity, storage, and compute. This week, I look at MaxLinear, Sandisk, and AMD. I believe that , these are the companies helping define the next phase of the AI buildout.

The original thesis strengthened as Mt. Holly completed its restart, Grundartangi returned close to full production, and Jamalco’s new turbine came online.

Dell's Q2 FY2027 revenue reached $47.0 Bn and non-GAAP EPS reached $7.04, beating pre-quarter consensus by about 4% and 43%, respectively.