Big Law Firms Likely Next Anthropic Deep Deal Prediction

The Freshfields × Anthropic deal announced April 23, 2026 defined a procurement tier that didn't exist publicly before — co-build with foundation model provider, 5,700 employees across 33 offices, +500% adoption in six weeks. The structural question for the next 12 months is which BigLaw firms fit the same template. Three criteria narrow the field: revenue tier above $2B, global office footprint with sophisticated AI risk-and-ethics function, and existing public AI deployment posture. This isn't a list of firms "in talks" — that would be fabrication. It's the structural read on which firms have the offer the foundation model provider would value, and which firms have the incentive to take the co-build track over the vendor track.

The three structural criteria for an Anthropic-deep co-build

Per the Freshfields deal terms, three criteria sat underneath the structural fit:

- Revenue tier above $2B annually. The co-build engineering investment is real — internal AI team plus dedicated lawyer time on co-design plus deployment infrastructure across multiple jurisdictions. Below the $2B revenue line, the engineering burden runs into 2-3% of annual revenue, which is hard to defend to a partnership. - Global office footprint with sophisticated AI risk-and-ethics function. A 33-office, 11-jurisdiction rollout requires a function that already does cross-border regulatory compliance for the firm's own technology decisions. Firms without an existing AI risk function typically can't absorb the rollout without political risk to the partnership. - Existing public AI deployment posture. The firm needs to have made some public commitment to AI tooling before the co-build conversation — pilots, vendor announcements, AI committee charter — so the co-build deal lands as the next step, not a from-zero pivot.

The second-order criterion: training-data quality at industrial scale. Anthropic gets value from co-build relationships because Magic Circle and AmLaw 50 work product is the highest-quality legal feedback signal available anywhere. A firm whose practice mix concentrates in one specialty has less to offer than a firm with depth across transactional, disputes, regulatory, and finance.

The third-order criterion: regulatory posture. Firms heavily regulated in their primary markets (UK Magic Circle, US AmLaw 50 with significant federal regulatory practice) bring jurisdictional complexity that genuinely improves the model. Domestic-only firms add less.

Latham Watkins — structural fit, open vendor question

Latham fits the criteria cleanly. Annual revenue above $5B, 33 offices globally, established AI committee with public posture. Per Latham's public technology coverage, the firm has piloted multiple AI tools without locking into a single primary vendor.

The structural read: Latham has the revenue and footprint to absorb a co-build. The firm's M&A practice depth alone provides training-data signal Anthropic would value. The open question is competitive — Latham's posture historically has been to evaluate multiple vendors before committing, which is structurally compatible with co-build but doesn't predict it.

The second-order read: if Latham goes co-build, the natural deal would mirror Freshfields' scope (firm-wide deployment plus co-development plus early access to future models). The natural alternative would be deeper Harvey or CoCounsel relationship — both compatible with the firm's existing posture.

Not in talks per public reporting. Structurally fits the template.

Kirkland Ellis — revenue tier match, posture less public

Kirkland's $8B+ annual revenue and global footprint are the largest in the AmLaw rankings. The firm fits the revenue and scale criteria with margin to spare.

Where the structural fit gets less clean: Kirkland's public AI deployment posture has been less prominent than Latham's or Freshfields'. The firm has confirmed AI tooling pilots but has been more conservative about public partnership announcements. That's not a disqualifier — it's a different go-to-market style.

The second-order read: if Kirkland goes Anthropic-deep, the deal likely lands without the press-release coverage Freshfields received. Kirkland's negotiating posture historically favors quieter relationships with deeper terms over public announcements. A co-build deal could be in place months before public confirmation.

The third-order read: Kirkland's PE practice depth — the firm's M&A and private equity work concentrates in a specific transactional pattern that produces highly structured legal output. That's training-data gold for any foundation model trying to improve transactional output. The structural offer is real even if the public posture suggests a slower negotiation cycle.

Not in talks per public reporting. Structurally fits the template with a different go-to-market profile than Latham.

DLA Piper — global footprint match, technology posture aligned

DLA Piper sits in a structurally interesting position: $3B+ annual revenue, 80+ offices globally, and a public technology posture that has historically been ahead of peer firms. Per DLA's published AI initiatives, the firm has made multiple public commitments to AI tooling and innovation infrastructure over the past five years.

The structural fit. DLA's office footprint is wider than Freshfields' (80+ vs 33), which means the deployment complexity is higher but so is the jurisdictional training-data signal. The firm's technology posture is the most public-facing in the criteria set — DLA has historically been willing to lead on technology adoption announcements.

The second-order read: DLA's practice mix is broader than Magic Circle peers, with significant volume in regulated industries (healthcare, financial services, life sciences) where compliance complexity drives genuine model improvements. That's a different training-data flavor than Freshfields' transactional concentration.

The third-order read: DLA going Anthropic-deep would likely surface as an explicitly public partnership rather than a quiet co-build. The firm's brand benefits from technology leadership announcements. If a co-build deal is in negotiation, expect press-release coverage closer to Freshfields' scale.

Not in talks per public reporting. Structurally fits and historically open to public announcement.

Allen Overy Shearman — Harvey relationship is structural friction

A&O Shearman is the partial counter-example. The firm fits the structural criteria — $3.5B+ revenue, 47 offices globally, sophisticated AI function — but the existing Harvey relationship running since February 2023 is structural friction against an Anthropic co-build.

The procurement reality: A&O built the original BigLaw AI flagship deal with Harvey. The firm's lawyers were trained on Harvey's interface. The procurement contracts have momentum. The internal advocates for AI deployment are Harvey-fluent. Pulling that into an Anthropic co-build means rebuilding the deployment surface and retraining the user base — engineering and political costs Freshfields didn't carry because Freshfields started fresh.

The second-order read: A&O is more likely to deepen the Harvey relationship than to pivot to Anthropic co-build. Harvey itself runs on Anthropic and OpenAI models, so A&O gets indirect Anthropic model improvements through Harvey's foundation model layer.

The third-order read: if A&O does go Anthropic-deep, it likely happens through a Harvey-Anthropic deeper integration rather than a direct A&O-Anthropic co-build. That's a different deal shape — vendor convergence with foundation model rather than firm-direct relationship.

Not in talks per public reporting. Structurally fits but with material switching costs from existing Harvey deployment.

Cleary Gottlieb — structural fit, less public AI posture

Cleary fits the revenue tier ($1.8-2B+ annual revenue) and global office footprint criteria. The firm's AI deployment posture has been less public than Freshfields, Latham, or DLA — Cleary has historically favored quieter vendor relationships and internal builds over partnership press releases.

The structural read: Cleary's transactional and regulatory practice depth in international markets produces training-data quality at the level Anthropic would value. The firm's posture toward public announcements is less aggressive than peers, which means a co-build deal could be in place without immediate market coverage.

The second-order read: per Law.com coverage, Cleary has been an early adopter of Thomson Reuters CoCounsel, which post-rebuild runs on Anthropic. The firm gets indirect Anthropic exposure through the rebuilt CoCounsel without a direct co-build. Whether that satisfies Cleary's strategic posture or pushes the firm toward a direct Anthropic relationship is the open question.

The third-order read: firms that go indirect through TR CoCounsel save procurement complexity but lose the early-access advantage Freshfields negotiated. Over a five-year horizon, the early-access advantage compounds materially. Cleary's posture decision in 2026-2027 is a leading indicator for how peer firms weigh that tradeoff.

Not in talks per public reporting. Structurally fits with a different vendor-relationship history than Freshfields.

What the next 12 months likely produce — pattern, not predictions

The structural argument: the next 12 months likely produce 2-3 more named co-build deals at Magic Circle or AmLaw 50 scale. The shape of those deals will define BigLaw AI procurement through 2028.

The pattern read, not specific firm predictions:

- At least one US-based firm will likely announce a deal of similar scope to Freshfields. The structural pressure is real because US firms competing for talent against Freshfields-trained associates need to offer comparable AI-tooling exposure. - At least one Magic Circle peer will likely follow Freshfields' template — Linklaters, Clifford Chance, or A&O Shearman by structural fit, with the latter constrained by the existing Harvey relationship. - At least one regional powerhouse outside the AmLaw 50 / Magic Circle frame may surprise — large Asian firms or specialized boutiques with concentrated practice depth (litigation, IP, restructuring) where the training-data signal is exceptional even at smaller revenue scale.

The second-order pattern. Anthropic's enterprise capacity for co-build deals isn't unlimited. The firm provides engineering and lawyer time to a finite number of co-development relationships. Once that capacity allocates in 2026-2027, late-entrant firms get the standard enterprise contract instead of co-build terms.

The third-order pattern. The firms that get co-build deals in this window will see model improvements first; firms on standard enterprise contracts will see them next; firms on consumer-tier or no Anthropic relationship will see them last. The compounding effect on competitive talent attraction over five years is real and likely understated. The Freshfields replication guide for mid-market covers what's available to firms outside the co-build window.

The Bottom Line: My take: Latham, Kirkland, DLA Piper, and Cleary fit the structural criteria for an Anthropic-deep co-build deal. A&O Shearman fits but carries material switching costs from the existing Harvey relationship. The next 12 months likely produce 2-3 more named deals at this scale; firms that don't move within that window get standard enterprise contracts instead of co-build terms when Anthropic's engineering capacity allocates. None of these firms are confirmed in talks per public reporting — this is structural fit analysis, not insider information. The firms with the cleanest fit and most public technology posture (DLA Piper, Latham) are the highest-probability next announcements.