IBM and Tesla report tomorrow. One already pre-announced a −7% Infrastructure miss; the other already filed a SpaceX chip-fab partnership.

IBM ($IBM) and Tesla ($TSLA) both report Wednesday, July 22. They are not the same story — but they share a week where AI physical spend is rearranging budgets and balance sheets.

IBM already told you the ugly part. On July 14 it furnished a CEO letter with preliminary Q2 results; the stock fell more than 20% in the session. Tomorrow’s call is less about “will they miss?” and more about whether the miss is timing (large deals slipped; Z wrap) or a durable budget reorder away from mainframe/transaction software toward servers, storage and memory.

Tesla already told you the volume part. On July 2 it filed Q2 production / deliveries / energy deployments. Tomorrow’s print is ASP, margin, energy profit, FSD/Robotaxi, and — the filing that matters for the SpaceX debate — what management says next about the SpaceX chip-fab partnership and the $2.0B SpaceX equity investment already on the Q1 cash-flow statement.

Research, not investment advice. Figures below are from SEC Exhibit 99.1 releases unless labeled otherwise.

Calendar (July 22)

We already wrote a same-day adjudication of IBM’s letter: IBM’s Q2 miss — enterprise Infrastructure vs AI physical capex. This preview goes deeper on what tomorrow has to confirm or break — and puts Tesla / SpaceX next to it.

---

IBM — the July 14 8-K that crushed the stock

What Krishna filed (preliminary Q2)

From the July 14 Exhibit 99.1 letter (accession 0000051143-26-000070):

Operating EPS bridge in the exhibit: GAAP $2.27 + acquisition-related $0.58 + retirement-related $0.08 = $2.93.

That mix is why the tape went violent. Headline growth nearly stalled. Operating EPS still grew. Segment mix told a story markets hated: the franchise investors paid up for (Z + transaction software) was the shortfall; the parts of IBM that look like “AI-adjacent kit and platform” were not.

Why the drawdown was about narrative, not just a miss

Context from the same letter and prior filings:

1. Q1 was strong. April 22 release: revenue $15.9B (+9% / +6% cc); Infrastructure +15% with IBM Z +51%; full-year guide affirmed — >5% constant-currency revenue growth and about +$1B YoY free cash flow.
2. Soft Infrastructure was telegraphed — at low-single digits. Krishna says April guidance already expected Infrastructure to decline low-single digits for the year, beginning Q2, wrapping the strongest mainframe-program start in IBM history (z17).
3. −7% was worse than that plan, driven by Z performance and the associated software stack, primarily Transaction Processing.
4. Late-June budget raid: clients “shift[ed] their quarterly capex spend toward servers, storage, and memory” to secure supply-constrained kit ahead of expected price increases. IBM “anticipated some supply chain related impact” but “did not anticipate the magnitude of the capex reprioritization.”
5. Execution confession: “this quarter we faltered… numerous large deals failed to close on the timelines we expected, driving the majority of our shortfall.”
6. Cybersecurity distraction also cited as a client headwind.

A >20% one-day move after that cocktail is extreme for IBM. Extreme does not automatically mean “wrong.” It means the market repriced mainframe-cycle + software-growth optionality overnight. Tomorrow is when IBM either re-anchors that optionality with backlog / deferred closes / FY guide — or confirms the reprice.

The paradox inside the same letter (do not skip this)

So buyers did not stop buying IBM-adjacent physical infrastructure. They reordered which IBM stack got the last weeks of June budget — and the high-ASP Z / transaction-software complex lost. That is crowding-out inside the IT budget, not proof that “AI spending is fake” or that “enterprise software is dead.”

For the longer adjudication of that distinction, see the July 14 deep dive.

Lightwell and quantum — real, but not the miss

The letter also unveils Lightwell (a $5B commitment; >20,000 engineers; open-source vulnerability clearinghouse; bank early adopters named) and a quantum wafer-foundry letter of intent ($1B CHIPS + $1B IBM cash; >$10B quantum over five years). Keep these in a separate bin. They do not explain the Transaction Processing shortfall. They will get airtime on the call; do not let them drown the Z scorecard.

Scorecard for IBM tomorrow (July 22)

1. Finals vs prelims — does $17.2B / Infrastructure −7% / op. EPS $2.93 hold, or does the “slightly different” warning matter?
2. How much of the miss was deal slip vs demand? — backlog, book-to-bill, “closed in July” language.
3. Full-year guide — April’s >5% cc revenue and +$1B FCF still intact, narrowed, or cut?
4. Software quality — was +5% a Transaction Processing air pocket, or a broader software deceleration under Red Hat / Hashi / Confluent?
5. Distributed Infrastructure backlog — does the ~$500M Power/Storage backlog convert, proving the “AI physical kit” demand is real for IBM too?
6. Tone on capex reprioritization — one-quarter scramble, or a multi-quarter budget hierarchy where Z closes second?

Live multiples: IBM valuation.

---

Tesla — deliveries already known; SpaceX is the under-discussed filing

Q2 volume already on the SEC tape (July 2)

From Tesla’s Q2 2026 production / deliveries / deployments Exhibit 99.1 (accession 0001628280-26-046717):

Two tells before any P&L:

• Deliveries exceeded production — inventory draw, not a production-only story.
• Energy storage deployments accelerated harder than vehicle deliveries. If Megapack margins hold, Energy can carry a larger share of the narrative than the auto tape alone.

Tesla itself warns in the same exhibit: deliveries and storage “should not be relied on as an indicator of quarterly financial results,” which depend on ASP, cost, FX, and the rest of the 10-Q. Fair. Still, volume is no longer the mystery for tomorrow — price/mix and opex are.

Last full financial print (Q1 2026)

From the April 22 Q1 update (accession 0001628280-26-026551):

Q1 Energy revenue was soft (−12%) even as the year-ago and Q4 run-rates were stronger — another reason tomorrow’s 13.5 GWh deployment print matters for sequencing Energy revenue back up.

The SpaceX potential — what is actually filed (not vibe)

Tesla’s Q1 update is unusually explicit. Three filed facts:

1. Partnership language: “Our partnership with SpaceX aims to build the largest chip fab ever: vertically integrating logic, memory and advanced packaging… This begins with the Tesla-owned Research Fab on our Gigafactory Texas campus.”
2. Product milestone next to it: final chip design of next-generation AI5 inference processor completed in April.
3. Balance-sheet / cash-flow fact: Q1 investing cash flow includes Purchase of SpaceX equity investment — $(2,002)M. Cash bridge commentary: the quarter’s cash change reflects $1.4B FCF and $1.2B financing inflow, “partly offset by $2.0B for the SpaceX equity investment.”

That is the cleanest SEC-filed Tesla↔SpaceX bridge available today. It is not a claim that Tesla owns SpaceX, that Starlink revenue hits Tesla’s P&L, or that the fab is already a foundry peer to TSMC. It is a claim that:

• Tesla is putting real capital into SpaceX equity;
• management wants the Street to underwrite a joint semiconductor build aimed at AI inference / training silicon scarcity;
• the first physical node is a Tesla-owned research fab in Texas, not a press-release moonshot with no address.

SpaceX itself is now in the public-market conversation after its S-1 / IPO map (Starlink economics, xAI consolidation, supply-chain graph). For Tesla’s July 22 call, the SpaceX questions that are filing-rooted are:

1. Status of the Research Fab — tools, process node ambitions, timeline.
2. AI5 — where it ships first (vehicle inference, Robotaxi, Optimus, training cluster?).
3. Any further SpaceX equity / related-party disclosure beyond the $2.0B already booked.
4. How this interacts with Dojo 3 custom silicon and the Cortex GPU clusters Tesla already lists as capacity ramps.

Separately, Tesla’s autonomy / robotics stack (FSD Supervised, Robotaxi ramp, Cybercab / Semi line prep, Optimus “ahead of mass production”) is the other half of the AI story. Those are Tesla products. The SpaceX fab partnership is the input scarcity hedge — chips — sitting underneath them.

Scorecard for Tesla tomorrow (July 22)

1. Automotive ASP and gross margin — can GM hold near Q1’s high-teens / ~21% zone while volumes jump ~25% YoY?
2. Energy — does 13.5 GWh convert into a revenue and margin rebound after Q1’s −12% Energy revenue?
3. Opex vs AI build — Robotaxi / Optimus / AI compute spend vs operating margin.
4. SpaceX — any incremental color on the fab partnership and equity stake.
5. FSD / Robotaxi KPIs — miles, geofence, paid miles, regulatory path.
6. Cash — FCF after another quarter of capex and any further strategic investments.

Live multiples: TSLA valuation.

---

One frame for the shared Wednesday

IBM is a budget hierarchy story. Tesla is a volume-plus-AI-capex story with a rare, filed SpaceX semiconductor overlay. Same AI cycle. Different place in the stack.

Stress-test assumptions in the DCF calculator.

Primary sources

IBM — July 14, 2026 Exhibit 99.1 letter (accession 0000051143-26-000070); Exhibit 99.2 non-GAAP; Q1 2026 Exhibit 99.1 (accession 0000051143-26-000036); prior deep dive: IBM Q2 prelim Infrastructure vs AI capex.

Tesla — Q2 2026 production/deliveries Exhibit 99.1 (accession 0001628280-26-046717, Jul 2, 2026); Q2 2025 deliveries (YoY base) (accession 0001628280-25-033842); Q1 2026 Update Exhibit 99.1 (accession 0001628280-26-026551) — SpaceX fab partnership + $2.0B equity investment; Q4/FY2025 Update (accession 0001628280-26-003837).

SpaceX public-filing context: SpaceX S-1 supply-chain follow-up.

Research, not investment advice.