The Watson Weekly is sponsored by Avalara - the agentic AI platform automating global tax and compliance for leading eCommerce brands.

Ryan Cohen made his move Sunday. GameStop, a $12B retailer, bid $55.5B for eBay, a $46B marketplace, at $125 per share in a 50/50 cash-and-stock split.

The 46% premium GameStop is leading with runs off eBay's February 4 close, the day Cohen started accumulating eBay shares. Shareholders looking at Friday's close see a 20% premium. Both numbers are real. Only one is going to drive a board response.

The cost case is the centerpiece of the pitch. $2B of annualized cost out in twelve months: $1.2B from sales and marketing, $300M from product development, $500M from G&A. The S&M cut is the load-bearing claim. eBay spent $2.4B in fiscal 2025 and added 1M net active buyers on a base of 134M. Under 0.75% growth. Cohen's read is that the marginal dollar of marketing spend is producing nothing, and he is probably right. He is making a much bigger claim about the average dollar, and that one is harder. Brand spend that looks wasteful in a steady state often shows up as customer attrition the quarter you turn it off. Anyone who ran a DTC P&L through 2023 watched it happen.

The EPS bridge from $4.26 to $7.79 assumes every cost dollar comes out clean and revenue holds flat. Both assumptions deserve a skeptical diligence team.

The piece of the deck with real operational substance is the store footprint. 1,600 US locations: on-the-spot authentication and intake, fulfillment nodes, and broadcasting studios for live commerce. eBay's authentication program is genuinely working in sneakers, watches, and graded cards. GameStop staff already grade hardware and cards every day. Whatnot is taking real share from eBay in collectibles live, and eBay has talked about live commerce for years without making it stick. A national physical network is something eBay would never build from scratch. This is the only claim in the proposal, I think, that survives an honest pressure test.

Financing pencils out closer than the WSJ coverage suggests. $9.4B in cash plus a $20B highly confident letter from TD Securities is $29.4B, against a roughly $27.75B cash leg. The stock leg is GameStop paper, which is its own conversation. Outside capital, possibly from Middle Eastern sovereign wealth, would be working on the equity side, not plugging a debt gap.

eBay's board now weighs Cohen's plan against the one already running. Q1 GMV up 18%. A 6.5% workforce reduction was announced in February. AI tooling is moving through the buying and selling flows. Bernstein's read this morning is that the turnaround is working and disruption is the wrong call.

I think the turnaround is working. I also think Cohen is right that the ceiling on this business is higher than current management is pricing in. The proposal does not yet show how to get from one to the other without breaking the thing eBay has finally fixed.

OpenAI putting up $1.5B for DeployCo, guaranteeing PE partners a 17.5% minimum return, while reportedly losing close to $14B this year? That is not a sales motion. That is a financing vehicle disguised as a sales pitch.

We have seen this movie before. Lucent and Nortel ran the same playbook in the late 90s, lending billions to customers so those same customers could buy gear back from them. It booked beautifully on the income statement. Then 2001 hit. Lucent posted a $16B loss. Nortel later paid $35M to settle SEC fraud charges. Lucent stock went from $84 to 76 cents.

The question for anyone underwriting the AI category right now is simple. What is the renewal rate on revenue that was not subsidized? What is the net dollar retention on customers who paid full freight? Because if a Big 4 consultant got a kickback to recommend Gemini, and a PE-owned portco was told from above to deploy Claude, you are not measuring demand. I'm seeing the distribution muscle, but it's being presented as ARR.

Show me the unsubsidized cohort. Then we can talk valuation.

If you are underwriting AI at current multiples, ARR is the wrong number to anchor on. The number that matters is the renewal rate on the cohort that paid full price, with no JV behind it and no consultant kickback in front of it. That cohort is the actual business. Everything else is sales and marketing spend booked on the top line, and we will not know which is which until these companies start disclosing the split. Operators have seen this pattern before. It does not end with a soft landing. It ends with a writedown and an SEC settlement five years later. Ask the question now while the answer is still cheap.

Tuesday, May 5

Wednesday, May 6

Wednesday, May 6

Watson Events & Webinars